What next for my 2.4c?

[Abecedaria]

Hey all,

I've just started work on a 2.4c as my first serious attempt at OCing. :) I've been able to boot at 3.0 @ FSB 250 but Prime95 gives a rounding error when the temps hit about 50c on stock Intel cooling.
On my MSI 865PE, I've got the AGP/PCI locked and Vcore is 1.575. (more voltage just makes it hotter)
RAM is running async @ 333(200FSB)/400(250FSB) in order to rule out the cheap generic RAM (2.5-4-4-8!) as a limiting factor.

My question is....where should i go next with this rig to get the most performance boost? Should I buy some nice PC4000 RAM and crank up the DRAM frequency or should I go for better cooling and try for a higher overclock?

I'm looking at a Thermalright XP-90 for cooling and 1Gb (2x512) geil PC4000 for the RAM.

Cheers,
abc

[hyperasus]

Yes buy Ram it will do you the most good. I have a 2.4c that overclocks quite nicely to 3.5GHz(on water); but it needs good ram because the low multi on that 2.4c will allow high FSB.

[fAlCoNNiAn]

Def. get better ram so that it might not be a problem. If you are looking for better cooling, look at the XP-120 by thermalright, should be good enough. make sure that you also have active cooling on the northbridge, since increase in fsb will make the nb get hotter as well. i suggest maybe putting some artic silver on the nb also.

but honestly speaking: if you do upgrade all of that, then probably you are going to get about *MAYBE* another 20 mhz fsb increase. so i guess this would be a great time to decide if you really need to spend your money on better parts for your current system or stick with what you have currently and save your money and get a better system in the end.

my 2 cents. :shrug:

[SkyFire47]

go for both. cooling's def. important cuz the ram won't matter if you can't keep your temps stable at your highest stable overclock attainable. minimum you could do would be to put AS5 with stock Intel cooler, but a third party heatsink would work best.

the right ram can take you a good distance. i was at 3.0 (250 fsb) and an upgrade in ram sent me to 3.4ghz (285fsb) on air.

[Elif Tymes]

RAMS pretty cheap right now, you could easily pick up a gig of 4000 for ~180 or so.

Add a Thermal Right XP-90 and you could prolly hit 3.2-3.3 Ghz with a ice Synced RAM/FSB

That would be about 230 dollars, and you could use the XP-90 in the future, and the 4000 RAM should have a decent resale value near the end of summer.

[Reefa_Madness]

Before you go out and buy ram consider trying to find your cpu's max fsb using the 3:2 divider which will cover you up to 300 fsb.

I am more inclined to believe that you would benefit most by upgrading the cooling from the stock Intel HS because this will allow you to increase the cpu's voltage which I believe you will need to do to get much over 250.

By testing using the 3:2 divider you can determine if in fact your cpu will run over 250. While temps will climb, you will be doing this for short periods of time and unlikely that you will reach, much less, exceed 70C or so. You won't hurt anything with these quick tests.

Once you've established that in fact you can exceed 250 fsb, I would think that the cooling should go hand in hand with any ram purchases that you make.

[Jimbob7]

Before you go out and buy ram consider trying to find your cpu's max fsb using the 3:2 divider which will cover you up to 300 fsb.

I am more inclined to believe that you would benefit most by upgrading the cooling from the stock Intel HS because this will allow you to increase the cpu's voltage which I believe you will need to do to get much over 250.

By testing using the 3:2 divider you can determine if in fact your cpu will run over 250. While temps will climb, you will be doing this for short periods of time and unlikely that you will reach, much less, exceed 70C or so. You won't hurt anything with these quick tests.

Once you've established that in fact you can exceed 250 fsb, I would think that the cooling should go hand in hand with any ram purchases that you make.

This is great advice. As I was reading down through the posts i was waiting for a reply like this.

[SkyFire47]

my computer wouldn't run 3:2 dividers with the particular ram i had. for me, this test would've proved invalid as my computer was rock solid at 3.0ghz but wouldn't go any higher no matter what dividers i tried with my old corsair ram.

the move to g.skill was the only way i saw my computer shoot past the "wall" Even now, my computer still won't run 3:2 dividers and that's with my gskill pc4800 rated ram.

hope this helps if you don't get too far with the change in dividers.

[Abecedaria]

Thanks for the awesome advice, everyone! Based on your feedback, I decided to find the max FSB for the 2.4c using the 3:2 divider and found some interesting details.
Using a Vcore of 1.7v, I was able to get it to post at 272 FSB, but XP wouldn't load. I dropped it to 267 FSB in order to get into XP, but Prime95 would give errors pretty quickly.
I ended up at 250 FSB with a 3:2 divider and a 1.7v Vcore to get a totally stable system. :rolleyes: I tried using the 5:4 divider (DDR400 with the loosest RAM timings), but it would crash fairly quickly in any game I loaded. The generic RAM I have in the system certainly doesn't seem up to the task, here. I was pretty frustrated at this point since I just didn't seem able to get any kind of decent OC out of this system.
Then I noticed something odd... SpeedFan was showing my 3.3v rail down at 3.14v with no load and 3.09v fully loaded. I decided to check my Vcore readings and found them at 1.7v with no load but dropping as far as 1.64v under load and averaging about 1.66v. Isn't this an excessive drop? I have a **cough** Rosewill 500W PSU, so that might be the problem. I also noticed my 12v rail is at 12.75v no matter the load, which seems high. Could the voltage regulator on the MoBo be the problem? Would any of these issues be causing the the poor OC?
I'm going to throw my stock 300W Delta PSU back in and see if it's more stable. If it is, I guess I'll have to RMA the Rose-swill for something better. :shrug:
Sorry for the novel. So here's my plan:
1) Test the PSU - RMA it if it's trash
2) The RAM is certainly holding me back, so I'll pick up nice PC4000 next
3) Loaded temps run around 45c at 1.5v Vcore and 58c at 1.7v Vcore, so I'll get the XP-90 last

What do you think? Hmmm, I guess I still have a bit to learn......

Cheers,
abc

[Abecedaria]

...one more thing... do you recommend hyper-threading on or off for overclocking?

Thanks,
abc

[ropey]

Hi and :welcome: to the forums.

First, to answer your question regarding Hyperthreading. I would leave it on as it is a valuable addition to the ableness of the system. I also think that running on a 1.7v is asking for trouble down the road.

Have you tested for stability at lesser voltages?

R

[Abecedaria]

Thanks for the welcome. This board is an awesome resource; I can't tell you how much info I've gathered just looking at the stickies.

As for the 1.7v vcore, I'm using it to find my max FSB and will drop it down once I figured it out.

Anyhow, I swapped in my old Delta PSU last night and I still have the same voltage problems. Under load, my vcore drops from 1.70v to 1.64v and then bounces around pretty wildly between the two extremes. I can't imagine that's too good for stability. :eh?: I've even noted this same behavior at stock voltage, 1.525v.

I assume that this means that the mobo's voltage regulation is rather poor. It probably also explains why the system will post as high as 275 FSB, but doesn't stabilize under Windows until about 250 FSB or less, even with a 3:2 memory divider. :mad:

Can anyone think of any other possibilities? Thoughts?????
abc

[ropey]

I ran my wife's P4 2.4C @ 3.4 with 1.57v. When I was in Windows there was a voltage fluctuation that ran between 1.5+ ~ 1.6+ volts. Through research I found that this was due to idiosyncracies with the i865/i875 chipset and performed a voltage droop mod on both my motherboards.

Regarding your MSI 865PE motherboard and the same fluctuations. Here is a thread with some information that could well help you in your quest.

http://www.ocforums.com/showthread.php?t=339171My voltage drops from about 1.58V in BIOS to 1.55V in Windows, to 1.52V under load. The drop under load is normal..... and it will fluctuate between 1.545 and 1.56V during idle if that's what you were wondering about.And here is the OCForums MSI i865pe search for voltage fluctuations that might as well be of benefit.

http://www.ocforums.com/search.php?searchid=810761

hth,

Ropey

[Daewood]

My MSI board undervolts by about .05 on the vcore and even when compensating for that I cant get it stable...

personally I think the MSI board is limiting your OC, if you pick up an IC7-G or a p4c800 you would be in alot better boat as far as overclocking goes

<--- will never buy another msi as long as I live...

[Abecedaria]

Thanks for the info, Ropey. (Everyone, really) Well, I finally have my 2.4c running at 3Ghz with a 1:1 memory ratio after upgrading my DRAM. (see my sig :D )

I still have the voltage fluctuation, but it seems a little less now. I looked though the info on the droop mod that Ropey linked and found another thread on the MSI web site: MSI vDroop Mod (http://forum.msi.com.tw/index.php?topic=59522.msg402199#msg402199)
This seems a little tricky as it requires soldering, I was hoping for a Bios based droop mod.....Any ideas for my mobo?

Temps are kind of high, so it looks like the XP-90 will be next...

abc

[ropey]

All droop mods are hardware based and as such need hardware intervention, thus I do not think you will find bios revisions bringing the fluctuation into stability.

You may wish to take the board out when you apply a new heatsink and fan assembly and have the droop mod performed at a tech shop. I would bring the relevent material to them when you do so if you do not feel comfortable in performing the modification yourself.

Good choice on the XP-90. I use the Zalman CNPS7000B-CU on the 2.4C @ 3.4 with 1.54v. This gives me 32 degrees idle and 41 full load. Considering the error of ASUS readings and adding the ~8 degrees differential this gives me ~ 40 idle and ~49 full load.

What is your Vcore set to at 3.0GHz?

R

[Abecedaria]

Yeah, I wanted to get the Zalman CNPS7000, but it won't fit. I figured the next best choice was the XP-90.



What is your Vcore set to at 3.0GHz?



I've had a hard time getting rock-solid stability at 3.0Ghz (250FSB), even with 1.7v! However, it's totally stable at 2.94Ghz (245FSB) with 1.6125v which is why I think I need the droop mod. I'm hoping the mod will stabilize my vcore, which swings around wildly right now, and allow me to get good stability above 3.0Ghz.
As I said, I can get it to post as high as 272FSB @ 1.7v, but it doesn't make it into Windows.

abc

[ropey]

I have never been impressed with MSI boards for overclockability and it seems to me that this might well be the problem with your overclock.

Quite possibly you are at a wall at 3.0C with that mobo. Certainly you need to pump a lot of voltage through that processor and heat/volts are the main cause of Sudden Northwood Death Syndrome.

R

[ojibewa]

I run my 2.4c M0 @ 3G all day long...

great CPU...

[Abecedaria]

I have never been impressed with MSI boards for overclockability and it seems to me that this might well be the problem with your overclock.

Quite possibly you are at a wall at 3.0C with that mobo. Certainly you need to pump a lot of voltage through that processor and heat/volts are the main cause of Sudden Northwood Death Syndrome.

R

Yeah, I probably wouldn't get an MSI board again. I took a chance on this one because it "looks good on paper". It has tons of overclocking options, RAID, GIGA LAN, you name it.... it just doesn't happen to be very stable. :bang head

I'll probably stick with ASUS or Abit next time. (or possibly DFI)

abc

[Jakalwarrior]

Sounds like yours does the same thing as mine. For me 248 at 1.57 turned out to be the best stable compromise for me though it isnt really dual prime stable for long periods. I have a hard time believing its my board or some other piece since more Vcore allows me to last longer in prime at higher fsbs or even become stable with large amounts (like 1.75 for 255+). I think intel just started putting out bum chips ;)

I am annoyed I was stupid enough to buy PC4400 with this setup though and the FSB only goes to 250.

[ropey]

Sounds like yours does the same thing as mine. For me 248 at 1.57 turned out to be the best stable compromise for me though it isnt really dual prime stable for long periods. I have a hard time believing its my board or some other piece since more Vcore allows me to last longer in prime at higher fsbs or even become stable with large amounts (like 1.75 for 255+). I think intel just started putting out bum chips ;)

I am annoyed I was stupid enough to buy PC4400 with this setup though and the FSB only goes to 250.
If you try your chip in an ASUS P4P800-SE you might find some really good movement with that chip and ram. I think MSI puts out bum boards for overclockers. I have tried the same chip in a MSI865PE Neo2 and a P4P800-Deluxe and needed much more vcore to get moving on the Neo2 and never did get the movement that the ASUS gave me. 2.4C @ 2.9 and on the ASUS I made 2.4C @ 3.4GHz (Same Chip). I just bought another 2.4 from a fellow here and made the same speed with it on the ASUS. Check my sig. Corsair 4400C25 gives me 1:1 also with dual prime for 24 hours. Superpi @ 32m and 3DMark Nature for 100 loops.

Now here's the kicker. I had to go past 1.65v on the Neo2 and I run at 1.53 at 3.4GHz on the ASUS. Have been running it at 3.4 for close to a year and eight months although I just moved to 1:1 lately as before it was running at 5:4....

Now maybe for higher multi procs like 2.8 and above the MSI's do better but with the lower multi procs they sure seem to need a lot of voltage to pump up the jam. This has been my experience at any rate.

R

[Jakalwarrior]

That does give a little hope but im still hurting from buying the ram
Im actually on an IS7 but it was a newegg refurb.