Ad
News

AMD beats Intel on desktop shipments.

Published on May 04, 2004

Figures from market research companies agree that AMD has started to bite into Intel's desktop market share, but to what extent depends on which set of analysts you choose to believe. Read more

Intel seeks to revive clone NB market

Published on July 14, 2005

Intel is seeking to revise a program promoting tighter cooperation between makers and distributors in an attempt to revive the clone notebook market, channel sources said. Read more

Rambus setback seen in patent infringement suit

Published on December 22, 2004

Rambus Inc. Read more

U.S. Government Gets Sued Over Secret IP Pact

Published on September 18, 2008

The word Conspiracy and Government tend to go hand-in-hand for most people. Read more

Latest Reviews & Articles

Highpoint Takes On Adaptec's SAS Controller

Published on December 19, 2008

Professional hardware can be expensive, but are cost effective business products really an alternative? We compared Highpoint’s new RocketRAID 2640X4 SAS RAID controller with Adaptec’s 5405 entry level card. Read more

GeForce GTX 295 Performance: Previewed

Published on December 18, 2008

After a frustrating second half of 2008, Nvidia is looking to start the new year off by reclaiming its single-card performance crown. We got our hands on an engineering sample GeForce GTX 295 to give you a taste of what you can expect in two weeks. Read more

Phenom Recycled: Athlon X2 7000-Series

Published on December 15, 2008

Just a couple of weeks before the introduction of its 45 nm Phenom II, AMD introduces a new dual-core chip. The Athlon X2 7000-series is basically a 65 nm Phenom with two active cores, but with the full L2 and L3 cache memory. Read more

Does Saving Power Mean Hurting Performance?

Published on December 15, 2008

Modern processors are capable of switching into power-efficient modes to save power when they’re idle, and an increasing number of motherboards offer dynamic features for the same purpose. Yet, the benefits come at a price. Read more

  Tom's Hardware Forums » Overclocking » General Discussions » Molly takes a bite at O/C ... seeks Divine hand !!
 

Molly takes a bite at O/C ... seeks Divine hand !!




Word :   Username :  
 
Bottom
Author
 Thread : Molly takes a bite at O/C ... seeks Divine hand !!
 
Profile: stranger
More Information

Hello --

Ok guys I have a Q6600 and 4 gigs (4 x 1gb) of DDR2-800 RAM.
I would appreciate some nods of acceptance or boots in the pants. I'm a little nervous but have read a lot and want to try my best.

I have the following system (specs at bottom) up and running. From all of my research, it seems that I can get an o/c to 2.8 on Q6600 without blinking an eye, so I'd just like to start there. You can tell by my questions that I am just a learner, but that's what I'm here for, to try and learn.

I'd like a comment on my plan to get to 2.8 (Q6600 G0). Have a whack at me if I'm wrong. Here is the important stuff. Remember I just want to grab 2.8 right out of the gate as a safe starting point...

CURRENT: [Using RealTemp 2.70] cold is 29 C, hot is 47 C, watching all 4 cores with Prime95 for an hour

CURRENT: Q6600 G0 @ 2448 MHz (stock) . That is 9x 272.
DESIRED: 2.800 MHz . Please confirm me: So I will change that to 9x 311. Yes? And do I leave CPU:DRAM ratio in BIOS at 1:2.00 -- which is where I assume it is now? (Actually it is on 'Auto' but I can discern a 1:2 ratio from real & effective clock rates in Everest)

CURRENT: 4x1GB Patriot PDC22G6400LLK 2.2V (runs 2.18V) with 5-5-5-16 timings
DESIRED: According to manufacturer, I should shoot for a lower timings settings of latency of 4-4-4-12. Would you say, good idea or bad idea? The 5-5-5-16 is what it gives me for "Load Optimized Settings" for DDR2-816 (also stock).

Thanks Guys !!! And below is the rest of the gathered info for the build.


This is a guts build, no case. The temps on all four cores (RealTemp 2.70) start out at 29-26-22-22 and after an hour of Prime95 Torture Test get to 47-46-42-42.


guts build, not in case yet...
IP35 Pro XE (bios 11) -- Q6600 G0 @ 9x272= 2448 MHz -- ZeroTherm BTF90 cooler -- 4x1GB Patriot PDC22G6400LLK 2.2V (runs 2.18V) with 5-5-5-16
8800GTS G80 320mb (GPU not o/c )
500GB Barracuda 7200.11 with the 32mb cache enabled

The CPU VID is 1.2500V (runs 1.21 V)
The CPU Speed is 9x 272 = 2448.0
Memory Bus is at 408.0 MHz
DRAM:FSB is 12:8

FSB Properties: clock as above=272 MHz
Real Clock: 408 MHz (DDR)
12:8 ratio

Related Product

Register or log in to remove.

Profile: member
More Information

Yea, what you have is fine. But 47 C is really good, thats about 30 C under the max temp, so as long as your temps don't go above 70 C in realtemp you're fine.

You may as well go to 3.2ghz, e.g. 9x 356
Set your Ram to a divider of 1:1, you can worry about overclocking it later

Now if you want to overclock you need to increase the core voltage. I have a Q9450 which is a 45nm process, and im not really sure about 65nm (your Q6600) voltages. So what you need to do, is up it to 3.2ghz (or 2.8ghz, if you don't want to go above) then see if it boots. If it doesn't, go back to bios, and up the vcore 1 step, and repeat until it boots into windows. When it does, run prime95, and if i fails before 8 hours, then up the voltage again one step and repeat until its stable for at least 8 hours. You can go for 24 hours if you're OCD or something :)

"Hello" lied the politician
Profile: Faithful Poster
More Information

actually, to get the q6600 as high as 3ghz (333x9) doesn't usually require ANY voltage tweaks - of course if it refuses to boot, thats the first thing to try though...
and agree, set ram to 1:1

Profile: stranger
More Information

Thanks so much, guys. That was very helpful. I went ahead and did what you said, and it is working out fine. I will put the details all below for the future benefit of IP35 Pro and Q6600 G0 users.

I ran two copies of Orthos and used affinity changes in Task Manager, to set each copy of Orthos to run on two cores. This got all cores tasked at 100% in Core Temp program. No core ever went above 61 C, and no errors or halting in the known prime tests were detected. I ran it about an hour that way. I do understand that I need to run it overnight for more assurance.

Here is the info I used from you guys to set everything to where it is now:
IP35 Pro XE (bios 11)
Q6600 G0 stepping
4x1GB RAM - Patriot PDC22G6400LLK
Stock: 9 x 272 -----> O/c to 9 x 311 in BIOS
Stock: CPU:DRAM ratio 1.00:2.00 -----> changed to 1:1 in BIOS
Voltage: didn't touch anything except put voltage for RAM at 2.2 V - but I had done that even before this o/c because this is what the manufacturer says to use.
Performance Test 6.1: Score went from 1307 to 1439
Temps: Never ranged above 61 C on any core, after the o/c. I noticed just how much of an affect the ambient room temperature had on things, though. After the sun went down and this room cooled, the max temp slid down to 56 C.

Thanks again, guys!


Message edited by molly_001 on 10-20-2008 at 12:14:31 AM
Profile: addict
More Information

try for 333x9 next :) it won't be much harder than the settings you have right now.

get back to us on your progress


---------------
http://valid.canardpc.com/cache/banner/465527.png
There's a statistic or bench for everything.
Profile: Faithful Poster
More Information

i would say 350FSB@8 mutliplier (8 has shown more stability than 9 weirdly enough) or 400x7 multi.

just put your RAM at factory timings and volts then put the frequency to DOUBLE whatever your fsb is eg 350fsb=700Mhz RAM; if your mobo supports it just put it "sync" or a 1:1 "Divider" with fsb :)

Profile: stranger
More Information

Thank you gentlemen. I had a question below, before going to 9x 333.
So, let me relate my experience so far just as a starting point.

I have Abit IP35 Pro XE, with Q6600 G0, and 4 sticks of RAM. Each is 1GB Patriot 6400.
The only thing I have done with voltages (which I know nothing about) is to set the DDR2
voltage to 2.2v as described by RAM manufacturer.

My first o/c step was to take Q6600 to 2.8 GHz, which is 9 x 311.
At that same time, I set CPU/DRAM ratio to 1:1 .

So this action had me at 2808 MHz on the CPU and the RAM was running at DDR2-624.
I tested this configuration very thoroughly, not only Prime95 but also Orthos set to
running all four cores on 100%. I did many hours of testing in this manner, using Core Temp
and Real Temp, and using CPU-Z to keep an eye on all settings to make sure that they
were all as expected. Everything good, no errors, no issues, heat OK.

The temps here never got about 51C on an 8-hour run of four core Orthos. However
during daylight/daytime testing of the same settings, I did get up to 61C on Core 0.
I think the difference is simply there was some sunlight coming into the room and shining
on my guts, which are laid out and running on a table.

Since this first run of [ 2808 Mhz / 1:1 ] was successful, I decided this morning to change
ratio of CPU:DRAM to 1:1.20 ratio. Why? Well because I noticed my RAM is only running
at DDR2-624 in above configuration. So now I changed that ratio to 1:1.20 . This makes
my RAM now at DDR2-748.

So to recap, I am at the moment with [ 2808 / 1:1.20 ].

Here is my question. I would like now to go to [ 3000 / 1:1.20 ], using 9 x 333.
I don't think the 1:1.20 will give me any problem here, because that will make it DDR2-799 .
However, should I use a multiplier lower than 9 for this?


You will also notice I have not mentioned anything about RAM timings. This is mainly
because I don't know anything about this, however I do know I am currently at 5,5,5,16.

Again, thank you in advance !!


  Tom's Hardware Forums » Overclocking » General Discussions » Molly takes a bite at O/C ... seeks Divine hand !!

Go to:
 

Google Ads