new rig 3.0c@3450Ghz and higher - Your Ideas wanted!!! ***THANKS***

[psykik]

Problem - 3.0c boots into dos at 3.66Gz at default voltage, but runs stable at only 3450Ghz. Tried up to 1.7v at 3.6 got stuck at win xp screen. Any ideas for a higher O/C appreciated!!!

If the rig below was yours, and this was the situation you were faced with, what would u do to max the O/C???

Given that i have excellent ram, the ram does not post beyond 240 fsb at even 2.7v? Perhaps the cpu is the bottleneck?


ic7 mobo
1GB kingston pc4000 2 x 512 (rated to do ddr 500 at 2.6v)
in slots 1 and 3 on mobo
antec trupower 480W
p4 3Ghz 800 fsb
radeon 9800pro
retail cooler (would it make a diff to try another one?)

with the above hardware im able to only o/c to 3450Ghz stable. While a reasonable o/c in its own right, be nice to get the MAX!!

fsb 230MHz
N/B strap psb800
Dram 1-1
fixed cpu/agp/pci
fixed agp/pci 63/33

cpu 1.6v
ddr 2.6v
agp 1.55v

mem timings are relaxed 3 8 4 4. Could be tightened once highest o/c reached.

Any criticisms of the above or ideas from fellow o/c'rs out there would be appreciated to get the max from this rig!!

Thanks!

 

[Corporate Thug]

apparently, hyperX has some issues with 865/875 chipsets

 

[lchyi]

Originally posted by: shady06
apparently, hyperX has some issues with 865/875 chipsets

Yes, apparently HyperX isn't running up to it's potential on the higher end memory series. I have the PC2700 which was the first HyperX and I've seen it run 200FSB before. But the newer ones aren't even hitting rated speeds which is kind of troublesome. Um and I also wouldn't expect a higher overclock from that 3.0C. I'm pretty sure it doesn't get much higher than 3.5 or 3.6 maybe on a good chip.

 

[infinite012]

Originally posted by: lchyi

Originally posted by: shady06
apparently, hyperX has some issues with 865/875 chipsets

Yes, apparently HyperX isn't running up to it's potential on the higher end memory series. I have the PC2700 which was the first HyperX and I've seen it run 200FSB before. But the newer ones aren't even hitting rated speeds which is kind of troublesome. Um and I also wouldn't expect a higher overclock from that 3.0C. I'm pretty sure it doesn't get much higher than 3.5 or 3.6 maybe on a good chip.

What they said.

 

[psykik]

UGH, does this mean i should sell this hyperx and go with corsair?

thing is ive ive tried all ratios Dram 1-1, 5:4, 3:2 with no chance of a boot beyond 3.66

was figuring if ive tried cpu volts up to 1.7v, and nothing beyond 3.66 at all mem ratios, must be the cpu not the ram?

pls add your 2cents here!

thanks!

 

[Corporate Thug]

Originally posted by: psykik
UGH, does this mean i should sell this hyperx and go with corsair?

thing is ive ive tried all ratios Dram 1-1, 5:4, 3:2 with no chance of a boot beyond 3.66

was figuring if ive tried cpu volts up to 1.7v, and nothing beyond 3.66 at all mem ratios, must be the cpu not the ram?

pls add your 2cents here!

thanks!

yeah, that would be a good idea. look into OCZ as well

 

[stevejst]

I don't think anybody understands this properly.
You are running stable at 3.45 GHz but Ram "does not boot" at 240?

240 implies 3.6 GHz. So what that got to do with Ram? And what makes you think your processor can do 3.6 GHz?

Have you tried Prime95 at all?

I bet once you do, you will discover that your processor might not be stable at 3.45 GHz either.

If you back step by step and run Memtest86 and Prime95 you will see what fails first, Memtest (memory) ot Prime (CPU). I bet on CPU.

As far as HyperX having issues with 865/875 I can assure you that my HyperX PC3200 (not 4000 though) has no issues with either one and will run as specified and even a bit better though I prefer GEIL any time.

 

[batmang]

heh, im running my 3500 hyper x at 231 / 466 ddr. no hiccups..... it aint the ram... also, run ur vdimm at 2.8v

 

[bjc112]

If you are running the 1:1 ratio @ 230 you should easily be able to pull 250FSB @ 5:4 2.8v...

Or am i missing something...

Reguardless, what are your temps? That chip may have hit the wall...

450mhz overclock isn't too shabby...

 

[psykik]

system temps: sys/ cpu/ pwm 35/48/38c idle

Update:

tried 2.8v on kingston hyperx pc4000, same problem persists:

CPU able to boot fine showing 3.6Ghz on post screen, but not able to go into windows at 3.6Ghz,
this is regardless of vcore 1.5 to 1.7v, 3.45Ghz seems max into windows and stable.

again 1-1, 5-4, and 3-2 ratios tried to no avail, at voltages up to 2.8v

ALSO!!

just tried one stick of corsair xms 3500 and still stable in windows at 3.45Ghz but no boot into windows at 3.6Ghz. Tried all mem ratios.

Ram it seems might not be the limiting factor here

Im thinking one of 3 things

1) better cooler - what is significantly better than retail to warrant additional cost?

2) mobo - using ic7 lastest 16 bios from abit, could a different mobo give better results?

3) have i hit the upper limit of what my cpu can boot at? 3.6Ghz?

What do you guys think?

thanks!

 

[oldfart]

What do you guys think?

3)

BTW, Prime95 will fail due to CPU or memory errors, not just CPU.

 

[sman789]

Originally posted by: oldfart

What do you guys think?

3)

BTW, Prime95 will fail due to CPU or memory errors, not just CPU.

what if ya set it to test the CPU mostly an ram barely....the newer version

 

[Slammy1]

Some of the newer Kingston does not run as quickly, different chips, but they are power hungry. I run 2.85V.

Almost certainly your CPU. A better cooloing solution (read extreme) will net you a bit more, but even at 3.45 you'll need more than the stock hsf and ttt. Run Prime95 to find what your stress temp is. I try to keep it below 55C, but I prefer to keep it below 50C early in the life cycle and 60C won't kill you. As someone mentioned, to find your max CPU speed change your ratio to 5:4 in order to eliminate the RAM as a possible problem.

 

[stevejst]

BTW, Prime95 will fail due to CPU or memory errors, not just CPU

True but I am assuming:

1. Testing first MemTest86 on boot, if it passes, memory is OK, then in Windows test Prime95, if it fails it should be clear the CPU fails first.

2. If it fails on boot on Memtest86, that is memory (or cache of the processor, that is possible). I passed Prime95 also while Memtest86 showed errors, just to note.

3. If it passes both and 3DMark fails, that is clearly the stability of the system.

4. Passed all? Up the FSB and repeat all from #1 again.

Not hard to differentiate. If you don't test methodically step by step you cannot conclude anything.
That is his problem since he mentions "stability" and "booting problems" but does not say whether he actually tested anything.
From booting in Windows/DOS you will learn nothing, you need to test step by step, else we here can all keep on guessing what the problem is.

 

[psykik]

stevejst - sorry wasnt so rigorous with testing but been playing UT for 3 hours with all candy on just great at 3450.

Thanks for the list 1-4, i will follow it tomorrow and report results.

Any ideas in meantime appreciated.

thanks

 

[stevejst]

There are many reasons why the system fails at some point but 3.45GHz with 3.0C is nothing to sneeze at, that beats just about anything available on the market today.

 

[WarCon]

Might just be your barrier. The IC7 is getting a much needed BIOS upgrade soon (the abit guy in their forum says late this week). That might help.

Disable the last two settings in the GAT section. And some people manually pick normal or enhanced for the second setting in GAT and leave the other 2 on auto. None of those have addressed my limitation, as I am hitting that 250-255 barrier (I hope). That area is the one specifically addressed, but that might net some kind of gains for everyone. It also fixes the incompatibility issue with a certain kind of Maxtor SATA drive.

Definately run Memtest-86 to see if some memory setting is why you lose stability (what are you judging stability by?). I don't think I would try to find my data transfer stability point (cpu<>northbridge<>memory & hard drive) with a game that could corrupt your install, when it won't run 1 sec of Prime95. Thats pretty unstable, but stable enough to boot. I dread that area and a bit more unstable as they are the data corruptor's.

I always try to boot up on memtest-86 after an increase in any performance setting or fsb as I think if its stable enough to run memtest-86 without error, then its less likely to corrupt files if I boot into Windows. (Yeah, I sometimes just wanna go benchmark at something I wish was stable - but try not too very often)

 

[SinfulWeeper]

Well, it sounds like you have a CPU limitation. This can be by one of three things.

1. The CPU just can not do it.
2. You may need a BIOS update (update the BIOS @ stock speeds).
3. Cooling on it is pretty lame.

The first one I can not help you with.
The second one is self explanitory.
For the third, I can give some advise. I like the Swifttch MCX4000 cooler along with a 92mm Vantec Stealth fan. For the fan you can choose a different one, that's just my preference. Also, on all my P4's I remove the IHS (removal voids your warranty, so do so @ your own risk).
Doing that always improved my OC speeds with anywere from 4-11mhz fsb stabily.

One other thing just popped in my head. Some of the newer memory out there only likes there stock timings. Try increasing your memory timings to what it's rated for... unless that is the rated cas timings.

 

[Overkast]

Originally posted by: psykik
system temps: sys/ cpu/ pwm 35/48/38c idle

1) better cooler - what is significantly better than retail to warrant additional cost?

Go for the SLK-900U Heatsink. It's huge, but with a good fan mounted, it conducts oodles of heat. It's one of the best HS out there. Just make sure it will fit in your case because it's so big.

Lots of people recommend mounting the Vantec Stealth 92mm fan on it because it's a quiet fan but moves a good amount of air... but I don't care about noise, so I'm all for the 92mm Vantec Tornado (it moves 30% more air at it's top speed).