Northwood Dillema - Am I Seriously this Unlucky?

[lastig21]

I got my second 1.6a in today, and I can't hit 133bus on this chip either. My last chip would load into windows, but was so slow I couldn't do anything with it.

My first chip was a Costa Rica that I got from GoogleGear, and I could only hit 2ghz stable with it. Assuming I just had bad luck, i ordered another cpu from mwave. This chip is a malaysia and doesn't run as well as the costa rica chip. I'm using the stock heatsink and thermal pad with this cpu, and I used the stock heatsink with as2 on the costa rica. I have my rdram multiplier set to 3 as it can't hit 133 bus, and have locked my pci bus to 33mhz.

My setup is:
Abit TH7-II Raid
Costa Rica or Malay 1.6a
2x256 megs Rdram running less than 400mhz


Possible Problems
My vcore is very low on this board. At the default voltage, I only read 1.42v, and never regain that .08v I'm missing. To run the chip at close to 1.50v, I have to set the bios to 1.575, and even that is a little low. All of my volt readings are about that low.

My cpu temps are outrageous as well. Temps are now fixed. 45C idle

I have an antec 400w psu.

I only have the 400mhz clock generators on the board. ICS 9212-03. I am currently trying to run the memory in spec at less than 400mhz (3*125 = 375). This is the absolute maximum that both chips can achieve. Because of this, I now think there is something else in the system holding me back.

Do you think I am just unlucky by not being able to achieve 133fsb with these chips, or is there something else holding me back. Please help me out. These chips need to run faster than what they are doing.

 

[BarneyFife]

Are you sure you have your hsf set properly? I got my 1.6a from mwave today and am currently running at 2.1 It sounds like a possible heating or motherboard problem.

 

[Viper GTS]

<< My cpu temps are outrageous as well. At idle with this stock heatsink, my temps are close to 60C. >>



From the little I've seen on NW overclocking, those temps are WAY too high. Are you sure you have the HS installed correctly?

Viper GTS

 

[Uclagamer_99]

hmm i picked up a 1.8A ( i know these don't overclock as well) but it wouldn't even hit 125FSB...quite disappointing...oh well

 

[lastig21]

I reset the heatsink and my temps have fallen to where they should be (about 45C idle). I still can't load the computer at 133fsb. If I give the cpu all the voltage I can Windows 2000 errors out. I get the "can't find ntoskrnl.exe". Do you think there is a chance my ram is crapping out on my at the higher clock speeds? I have the multiplier changed so the ram is running at 400mhz, but do you think it might not be able to run 133x3. I still haven't ruled out the motherboard as being my problem causer.

 

[Nate420]

Glad to hear ya got the temp under control. I would say it's either mobo or ram. From the results of everyone else, and my own, I can't see 2 different P4s that wouldn't do 133fsb. Mobo would be my guess, but what the hell do I know.

 

[Durron]

I'm having similar problems with my 1.8A and TH7II

I can only get it totally stable at about 2ghz, which is pretty disappointing...no matter what kind of settings I put on the memory.

 

[Duvie]

Mobos vary on ocing of this northwoods....

1.6a...majority get to 150fsb....2400mhz

1.8a...majority get to 133fsb...2400mhz...

2.0a's rarely get to the 133fsb promise land...generally around 120fsb for again 2400mhz

SEE THE CONNECTION!!!!

There are the top ocers getting 2.6 but they are few and far between...

there are no guarantees in ocing....BUt you can do the research get the best mobo with options, best power supply for delivering the juice, and the best ram to take the speed...In the end we can't pick the exact chip and there is the wildcard...

 

[Nate420]

Out of curiosity, what was up with the HSF?

 

[Uclagamer_99]

but anyways i think the OC is due to the chips themselves and not your mobo

i had a 1.6A that would boot up to 145FSB
i just picked up a 1.8A and tested it in the same mobo and it would only boot at 124FSB, same memory and everything was used

so the overclockability of your chip is definitely luck of the draw...it would be odd however that intel would not test these chips and then sell them at the highter pricepoints as 2.0A or 2.2As...usually when testing wouldn't they sell the chips that fail to run at the higher clock speeds as 1.6As and 1.8As? but this would only be weird if we were making the assumption that all northwoods are the same just locked at different multipliers

 

[lastig21]

The HSF I guess was improperly mounted. I took it off, reset it on the cpu and clamped it down. I don't know what I did differently, it just seemed to work the second time.

I am going to hope that these chips will still do 133fsb. Hopefully when my new memory comes in I will be able to check for sure. From the looks of it, I think the costa rica chip might actually go farther then the malaysia. I can only run the chips at 2ghz right now, though the malay needs 1.5v while the costa rica was happy less than that.

If anybody lives in the Houston area with a P4 setup capable of hitting 133 fsb, I would like to test these cpu's in that system. I could throw in a blue cold cathode light and pick up some lunch or something.

 

[manko]

<< so the overclockability of your chip is definitely luck of the draw...it would be odd however that intel would not test these chips and then sell them at the highter pricepoints as 2.0A or 2.2As...usually when testing wouldn't they sell the chips that fail to run at the higher clock speeds as 1.6As and 1.8As? but this would only be weird if we were making the assumption that all northwoods are the same just locked at different multipliers >>



Somewhere (probably H-OCP), a second hand account of an Intel insider said that certain 1.6A chips concievably could have passed the tests to sell at 2.2Ghz, but the sales/marketing folks needed more 1.6A's to fill demand, so they just marked them down. IF that's true, it would seem like certain batches from the same factory could be a bunch of marked down 2.2 chips, while another batch (or a batch from another country) could actually be only 1.6 rated chips (only passing the 1.6 tests). The other thing is, I'm sure Intel builds in a testing buffer, probably at least 200MHz...

 

[jiffylube1024]

<< hmm i picked up a 1.8A ( i know these don't overclock as well) but it wouldn't even hit 125FSB...quite disappointing...oh well >>





<< i just picked up a 1.8A and tested it in the same mobo and it would only boot at 124FSB, same memory and everything was used >>



Hmm, now I'm not alone anymore! I got my 1.8A a few weeks ago and I was the runt @ 125 FSB (1.60 volts - rock stable btw); can't even POST at 133. It seems like Intel is speed binning their CPU's now, as they definately weren't before! That's the only way I can explain a 1.6A not reaching 133 FSB (I had not heard of any account of that not happening before this post). Do notice that many people are running their o/ced chips at very high voltages: 1.75-1.8V and above with various techniques. Those with 1.6A's who can't reach 2.133 .. make sure you try 1.625 + volts, even 1.7 or 1.75, just to see if it will work at that high voltage.

Also, if you have Rambus memory, note it is very possible it's the memory holding you back as only high quality samsung memory seems to reach 133+ FSB. If you're running PC2100 DDR then it's obviously the chip holding you back.

 

[Uclagamer_99]

intel is binning there chips? mine was dated back to early january though before the 1.6As and 1.8As were becoming so popular...odd, i guess i doesn't really matter im happy running at mine at stock...i don't want to have a half-arsed overclock so i figure i should just stick with stability

 

[jiffylube1024]

I dunno if they are or not but it would seem to indicate they are if everyone is getting 2.133 GHz no problem and almost the majority are getting 2.2-2.4+ GHz, and now people can't get it to 2.13; same for the 1.8A.

 

[THUGSROOK]

ditch the RDram guys

 

[ginfest]

<< ditch the RDram guys >>



I wish I had a 1.8A running at 2.4G/136FSB/340DDR and Sandra mem scores of 2400+ rather than my 1.6A/2.1G/133FSB/PC1066 Rdram with Sandra mem scores of 3300+

 

[Duvie]

memory scores are more like 2600+ at that speed....thank you very much....Give us a 3:4 ratio asus and it will be more like 2800+ and cutting the difference...

Like via products I would sooner eat bugs then support that company (rambus)....My ddr makes me able to jump back into an amd setup if need be...At pc2700 that should support amd platforms for awhile...

Edit: It is unlikely intel gets such a fluctuation in yields of chips to say 1.6a's couldn't handle 2ghz or 2.2ghz...Just a downsize for market demand...Have you heard of very many 1.6's that can't do 125fsb or 1.8's that can't take 112fsb??? I haven't....

 

[Uclagamer_99]

Edit: It is unlikely intel gets such a fluctuation in yields of chips to say 1.6a's couldn't handle 2ghz or 2.2ghz...Just a downsize for market demand...Have you heard of very many 1.6's that can't do 125fsb or 1.8's that can't take 112fsb??? I haven't....

well you are making a big general statement in that you seem to be assuming that everyone who posts on a computer hardware forum is a good representation of the total population of people with these cpus but then again that could be the case


these p4 posts kinda remind me of the Visiontek GF3 Ti200s awhile back in November...it seemed like a majority of people got very high overclocks...or at least to GF3 speeds but at the same time there was definitely a miniority who could not overclock their card at all

 

[jiffylube1024]

<< these p4 posts kinda remind me of the Visiontek GF3 Ti200s awhile back in November...it seemed like a majority of people got very high overclocks...or at least to GF3 speeds but at the same time there was definitely a miniority who could not overclock their card at all >>



True that. My friend's MSI GF3 Ti200 goes to about (I think) 210/460+ core/mem (all I remember was it was a piddly o/c compared to what other people are getting). Mine goes to 230/520 without breaking a sweat. It seems luck is a very big part of it.

 

[KenAF]

The problem here....

Is that some people who go with RDRAM don't perform any research beforehand.

They don't buy the right RDRAM that will overclock. And they don't understand that the TH7-II has 400MHz clock generators, and only some of these boards (40% or 50%) are able to run the memory at full 533+MHz speed. They also don't understand that you must download and flash the latest bios in order to get increased cpu voltages, and the ability to keep the PCI and AGP bus in spec. You can download the latest bios right here.

 

[Flatbroke]

jiffylube1024, it is luck, in that the ram you got on the card determined how fast you could OC it. Visiontek early ti200, pre week 43, would not OC like the week 43 on up did with 5ns ram.

 

[lastig21]

Just checked the clock generators, ics 9212-03. I really thought these were capable of 1066 speeds, but they are only rated for 800. Looks like I'll need to check out a new motherboard if the new memory doesn't solve the problems. I can run either chip at 2ghz rock stable. One mhz more and I have problems. I no longer think it is the chips.

 

[kulki]

<< God gave men both a penis and a brain, but unfortunately not enough blood supply to run both at the same time >>


ROTFL

 

[Durron]

First off the rambus I'm using will run fine at 533. Second, the abit th7II allows you to use 3x instead of 4x on the ram so that kinda rules out the rambus problem. (at 133fsb and rambus at 3x it doesnt even full get back up to 400mhz on the ram). Also, you posted a link to a bios #'d h77 (link done wrong btw)...latest on abit's website is http://210.192.220.184/ftp/pub/bios/th7-ii/th7h38.exe.

At this point my 1.8A is only TOTALLY stable at around 2ghz. I can post all the way up to 2.4 (maybe more, havnt tried), but windows will crash. Clean installs of windows doesnt help, all PCI cards out doesnt help. I have a good quality power supply and good quality rambus. I figure it's either a crappy CPU or a crappy MB...hard to tell at this point.

And btw the th7II undervolts, at 1.5 i get 1.45 as the actual reading, if i set it to 1.625 (max allowed with h38 bios), it runs at about 1.57 actual...pretty lame.

Not to mention the CPU temp in the bios always reads around 45 degrees Celcius, even when i first turn it on after being off all night =/ Heatsink is barely warm to the touch, yet is attached correctly (reattached many times during various testing), using thermal compound. Hell so far it's gotten the same 45C readings using the TIM, no TIM or compound, generic compound, and ASII...same thing...something has to be wrong there.

Also if I overclock at all (i'm talking to like 1.9ghz even), most of the time when I reboot, the screen looks all f'd up, so i have to cold boot and it works fine. Doesnt do this at 1.8 ever. I'm going to try borrowing a 1.6A from work when we get some in and try it, if it sucks too I'm betting on the MB