Overclocked but ready for a little more..

thudo

n00b
Joined
Jul 19, 2006
Messages
29
I posted my purchasing specs here just under a month back so here I am: super happy but just needing some advice from the Bios and Hardware Masters here on Hardocp.

Firstly.. here are my specs on where we are now:

Code:
CPU Used: Conroe E6700 Retail B2
Motherboard: Asus P5W DH (1301 Bios)

My Current Settings:
AI Overclocking: Manual :D
CPU Frequency: 350
DRAM Frequency: DDR2-1050Mhz
Preformance Mode: Auto
PCI Express Frequency: Auto
PCI Clock Synchonization Mode: Auto

Memory Voltage: 2.20V
CPU VCore Voltage: 1.5000V
FSB Termination Voltage: 1.40V
MCH Chipset Voltage: 1.55V
ICH Chipset Voltage: Auto

RAM Used:
OCZ DDR2 PC2-8000 Platinum Extreme Edition Dual Channel 1000MHz DDR2 
Recommended: CL 4-5-4-15 2.2 Volts (use a VDIMM of 2.25V ± 5%)

My Current Settings:
Dram Cas# Latency: 4
Dram Ras# PreCharge: 5
Dram Ras# to Cas# Delay: 4
Dram Ras# Activate to Precharge: 15
Dram Write Recovery Time: 4

CPU temp @ 3.5Ghz (Idle 38c, Load: 41-42c)
Cooler: Scythe Mine + Arctic Silver 5 Thermal Paste
Motherboard temp @ 3.5Ghz (Idle 33c, Load ~35c)
Power Supply: OCZ GamerXtreme 700W

Now my questions:

1) After overclocking @ 3.20Ghz I noticed that when I try to restart the computer from WindowsXP the computer totally shuts down.. waits a second, then turns back on and proceeds as normal. It also does this when exiting from the bios. It never did this before under 3.20Ghz overclock. Should I be worried about this? I noticed when the computer comes on @ 3.20Ghz and up there is a slight draw of power on my lines (have a lamp connected to the same line as comp). I am switching this over to a 830W Backup unit so not worried. Note the PSU unit above.. Should be enough wattage..

2) And my main question is: With the above CPU and Memory.. Is there more room for both to improve? What should I be watching for above to push to 3.6Ghz (which is VERY high for a E6700)? How am I doing sofar with the above?

Thanks all.. Yes I am going to run benchmarks and tests to ensure it doesn't crash after 3-5 or more hours. Just getting a feel for the situation as I do not want to kill my specs. ;)

Thanks all.. comments and advice VERY WELCOME!

TuD.
 
Just an FYI..

With the above settings, I'm seeing errors in the 16M and 32M tests for the SuperPI tests as well as the Go-Orthos-Beta. I gather this is normal for my settings as I need to tweak them. Any thoughts everyone? Thanks!
 
Anyone can help? Just want this firmed up asap! Thanks HardOCP community!
 
Update...

Code:
CPU Used: Conroe E6700 Retail B2
Motherboard: Asus P5W DH (1301 Bios)

My Current Settings:
AI Overclocking: Manual
CPU Frequency: 366
DRAM Frequency: DDR2-732Mhz
Preformance Mode: Auto
PCI Express Frequency: 100
PCI Clock Synchonization Mode: 33.33Mhz

Memory Voltage: 2.20V
CPU VCore Voltage: 1.4875V
FSB Termination Voltage: 1.50V
MCH Chipset Voltage: 1.65V
ICH Chipset Voltage: Auto
Hyper Path 3: Disabled
Dram Throttling Threshold: Disabled

RAM Used:
OCZ DDR2 PC2-8000 Platinum Extreme Edition Dual Channel 1000MHz DDR2
Recommended: CL 4-5-4-15 2.2 Volts (use a VDIMM of 2.25V ± 5%)

My Current Settings:
Configure Dram Timing by SPD: Enabled
Dram Cas# Latency: -
Dram Ras# to Cas# Delay: -
Dram Ras# PreCharge: -
Dram Ras# Activate to Precharge: -
Dram Write Recovery Time: -

CPU temp @ 3.66Ghz (Idle 35c, Load: 41-45c)
Cooler: Scythe Mine + Arctic Silver 5 Thermal Paste
Motherboard temp @ 3.66Ghz (Idle 33c, Load ~35c)
Power Supply: OCZ GamerXtreme 700W
Note my new changes.. I'm now running @ 3.66Ghz with SPD enabled and some voltage changes..

Here's what works:

SuperPi = All Tests complete fine
Prime95 = All Tests (ie Torture Test) complete fine without errors (left for 1 hour)

And what doesn't:
Orthos Beta = Only Last CPU (Gromacs Core) test works perfectly.. any other test errors out in under 3 minutes.

Anyone understand whats going on here? Seems bizzaro to me.

Note: I am running WinXP + SP2 (all updates) but have dropped my Paging File down to 150mb (no need to have a large one anyway with 2gb ram). That shouldn't cause a prob with Orthos? Wouldn't think so.

Any other synthetic tests or bios configs to consider? Just need to know if I should just be ignoring this "ghost in the machine" Thanks everyone!
 
your ram voltage is set to 2.200 but later in the report it states your ram needs 2.25 volts. give it the extra .05 volts, bumping it up to 2.30 would even be better and it would still be withing the +- 5% tolerance.

set the memory timings to 5 5 5 12 in the bios until you determine your cpu wont go any higher. Then start running memtest86 and tighten up the memory timings until you start getting errors and back off a bit.

btw you are probally at the wall or darn close with that cpu on air amd hence the testing failues.

dont know why you did what you did to the paging file, dont, set it to let windows manage it or set to to 2GB.
 
I'll increase Ram to 2.30V

However, don't I want to keep Ram @ SPD so I can first troubleshoot the CPU first? No sense in doing both at the same time, no?

Not sure how I could be "at the cpu wall" when the CPU temps at load are far below what others are getting with serious OC'ing by air?

As for paging.. its disabled as XP doesn't need that much cache with that much physical ram. The cache is there for systems less than 2gb. Once its achieved or gone over, why a major paging file?
 
The Reboot problem is a P5W DH, mine does that as well when restarting. I also have to set my SPD times to auto to get past 355fsb (5-6-6-18 when they're supposed to be 4-4-4-12 @ 800mhz). And anything above maybe 404 FSB will not pass Orthos no matter how much voltage I give it, but games run fine for hires and have had no problems at all really with a [email protected].
 
Yeaa I figured as much: I'm getting to the point that Orthos-Beta is a "anally-retentive" torturer. ;) I tried some intensive computation apps I had a friend write for me a month back and it works flawlessly (almost 5x faster than on my old P4 3.0Ghz). Hell.. thinking of putting my DRAM frequence back to 1098 like it was (currently its 1:1 with the CPU/RAM @ 366:732) to get that insane ram speed boost. All games I've tested work but we'll see - I want to compromise as little as possible but system stability is still important in the long run.
 
The idea is to loosen up ram timings to take the ram out of the picture while you are determining the highest possible cpu overclock.

It depends on what is programmed in the SPD chip and how well those preset timings match what you are trying to do with the FSB. The main point being that if you dont know EXACTLY what is in there and at what frequencies the chip will automatically change the timings you have an unknown variable. There are programs that read that stuff and let you know so you can proceed with an understanding of what your memory is doing as you really push the machine. However setting the timing manuall to 5 5 5 18 or even 6 6 6 23 takes the ram completley out of the picture if your FSB/memory clock divider is set properly. This allows you push the cpu as [H]ard as you wish without worring about the ram. Then you go back and tweak the ram.

Either run the benchmark and beleive it or dont run it. Unless there is a documented issue with the benchmark there is a issue with the machine. I have never used that one and it sounds like a really good one, gonna find it and give it a try. My point is that when you really stress your setup and find it lacking (benchmark/test fails) instead of determining what the problem is you tend to want to ignore it. Hey, its your rig, do what you want.


" the cache is there for systems with less than 2 GB " please post a link where you read that crap, I would like to see it. Not going to spend the morning trying to explain the use of a system paging file in modern operating systems. do what you want with it, its probally not affecting your overclock either way. Depending on how the code in the test you are failing is designed, a too small page file might cause an issue, might not. /shrug

Good luck, gonna try that test on my machine and see if I can melt something. :D
 
I'll try manually setting ram to those loose settings. That also makes sense that setting it to AUTO might cause wonky settings in the background.

As for reducing the XP paging size down to 100mb or even off. I've heard that years and years ago. Makes sense since physical memory is now abundant. Sense I reduced the paging size down to 150mb I noticed no slowdown. In fact.. its speedy as shiet. Might also be the fact I have the new WD 10,000 150gb SATA drive but oh well.. ;)

Still.. yer logic is sound. I'll take that under advisement as I am always open to ideas. Thank you Bill!
 
Update - Ram PRIME95 for 12+ hours under same above specs.. No problems.. CPU Load Maxed @ 46c... Anyone else think Orthos is screwy? I do. ;)
 
Okie.. ongoing with my tests I have some questions to consider:

1) For Ram - should I have tighter timings (ie. 4-5-4-12) and keep Vdimm @ 2.20V or run by SPD and then increase DRAM Frequency much higher than where I am to 800+ then increase Vdimm higher than 2.20V

2) For this kind of ram, what is the usual "breaking point" for over-volting? 2.40V?

I am **VERY** satisfied with the current overclock @ 3.66 (1 full Ghz over stock) running @ idle: 35-36c load: 42-45c max. However..

When I first started this thread note my insane configuration - I know the ram was waaayy to high but the system actually ran well + the ram was insanely fast. I guess it comes down to what the ram can handle as its obvious the CPU can handle it pretty high. Thanks everyone.
 
Okie.. finalizing things..

Code:
CPU Used: Conroe E6700 Retail B2
Motherboard: Asus P5W DH (1301 Bios)

My Current Settings:
AI Overclocking: Manual
CPU Frequency: 366
DRAM Frequency: DDR2-915Mhz
Preformance Mode: Auto
PCI Express Frequency: 100
PCI Clock Synchonization Mode: 33.33Mhz

Memory Voltage: 2.25V
CPU VCore Voltage: 1.5000V
FSB Termination Voltage: 1.50V
MCH Chipset Voltage: 1.65V
ICH Chipset Voltage: Auto

RAM Used:
OCZ DDR2 PC2-8000 Platinum Extreme Edition Dual Channel 1000MHz DDR2 
Recommended: CL 4-5-4-15 2.2 Volts (use a VDIMM of 2.25V ± 5%)

My Current Settings:
Configure Dram Timing by SPD: Disabled
Dram Cas# Latency: 4
Dram Ras# to Cas# Delay: 4
Dram Ras# PreCharge: 4
Dram Ras# Activate to Precharge: 12
Dram Write Recovery Time: 4
HyperPath 3: Disabled
DRAM Throttling Threshold: Disabled

CPU temp @ 3.66Ghz (Idle 35-36c, Load: 41-45c)
Cooler: Scythe Mine + Arctic Silver 5 Thermal Paste
Motherboard temp @ 3.66Ghz (Idle 33c, Load ~35c)
Power Supply: OCZ GamerXtreme 700W
Now with that it doesn't pass Orthos yet but everything else does. (Well.. have't run incidences of Prime95 yet). I code Advanced Skirmish AI for Dawn of War/Winter Assault and have max details @ 1600x1200 with 5 HARD AIs and never have probs (game does not use dual core) so I need to try something like Max8.

Anyway.. I can get the computer stable @ 3.5 Ghz with 1:1 with the ram. I still want to have faster ram speeds but right now with the above its fast. Anyway.. coming to finally love the speeds. I could compromise with slightly slower ram speeds and make it true 1:1 @ 3.66Ghz then just tighten the mem timings even more. Comments?
 
Update: Think I have a winner..

Code:
o CPU Used: Conroe E6700 Retail B2
o Motherboard: Asus P5W DH (1301 Bios)
o Ram: OCZ DDR2 PC2-8000 Platinum Extreme Edition Dual Channel 1000MHz DDR2
Recommended: CL 4-5-4-15 2.2 Volts (use a VDIMM of 2.25V ± 5%)
o Power Supply: OCZ GamerXtreme 700W
o Video Card: XFX 7600gt @ 620/830

CPU temp @ 3.66Ghz (Idle 35-36c, Load: 41-45c)
Cooler: Scythe Mine + Arctic Silver 5 Thermal Paste
Motherboard temp @ 3.66Ghz (Idle 33c, Load ~35c)
Video temp: (Idle 41c, Load 47c)

My Current Settings:
AI Overclocking: Manual
CPU Frequency: 366
DRAM Frequency: DDR2-915Mhz
Preformance Mode: Auto
PCI Express Frequency: 100
PCI Clock Synchonization Mode: 33.33Mhz
Memory Voltage: 2.30V
CPU VCore Voltage: 1.5250V
FSB Termination Voltage: 1.50V
MCH Chipset Voltage: 1.65V
ICH Chipset Voltage: 1.20V

MicroCode Updation: Enabled
Max CPUID Value Limit: Disabled
Execute Disable Function: Disabled
Enhacned C1 Control: Disabled
CPU Internal Thermal Control: Auto
Virtualization Technology: Disabled
Intet SpeedStep Tech: Disabled

RAM Settings:
Configure Dram Timing by SPD: Disabled
Dram Cas# Latency: 4
Dram Ras# to Cas# Delay: 4
Dram Ras# PreCharge: 4
Dram Ras# Activate to Precharge: 12
Dram Write Recovery Time: 4
DRAM ECC Mode: Disabled
HyperPath 3: Disabled
DRAM Throttling Threshold: Disabled
I run various tests with a game I am testing and normally the game would die during the 1st round OR during the 2nd. Now it never dies with the above while keeping my original specs I want. Orthos still dies after 2mins but other benchmark apps work fine.

I might try increasing the Dram Write Recovery a little as a test to see what Orthos does. Not expecting much.
 
How are you getting the temps? They seem really low.

First off your MB (system) temp @ 33, perhaps it's quite cold in your room, but mine runs around 38 at idle

But the CPU temps is what I'm wondering about. I get 34c STOCK (e6700) with a Zalman 9500 and a nexus 120mm right above it. So, with your OC to 3.66 (from 2.66) seems AMAZING at only 1-2 degrees C over my stock setup with an Antec P180 case. As a base, I'm getting 41-42 CPU temp IDLE all day at 3.00ghz.

Yet your load temps are what is truely outstanding at 41c. I'm getting 65c with dual prime95 running. I don't see how we are nearly 20c apart AND how you only rise 6 or so degrees C from idle to full load.

Either we are measuring differently or something a bit 'funny' is going on here. Do you have some heavy duty fans just raging?

As a side note, and please forgive my inference but it doesn't do any good to offer too low of temps to look good, when it throws off the real curve.
 
Ditto gpm3! 100% agree - temps look a bit too good...

Thudo, what app are you using to measure your CPU temps? You may be using something that's not giving you an accurate result.

I use Intel's Thermal Analysis Tool (TAT) which is regarded by many here on the forum as the most accurate - especially on C2D CPUs. You can get it here: http://www.techpowerup.com/downloads/392

Core Temp is another one that is supposedly just as accurate. Get it here: http://www.thecoolest.zerobrains.com/CoreTemp/

Again, what are you using?
 
Back
Top