2.4C @ 3.0 GHz maximum overclock on IC7 w/ Zalman 7000. Did I get a runt or what?

[jiffylube1024]

I just got my 2.4C chip yesterday (brand new from a computer store in Canada) to go along with my IC7, gig of Twinmos PC3200, Zalman CNPS7000A-CU etc.

I first booted at normal, then went to 3.0 GHz at stock voltage (which undervolts to 1.47V on this board). Booted and ran fine for awhile. Then I tried 3.2 Ghz. No go. 3.12 GHz (260 FSB). Posts and get most of the way through booting but crashes before Windows. This is all with the memory at 3/2, so it is not limiting me.

So, I upped the voltage to 1.575 in the BIOS (1.54V actual). Exact same results. I even tried 3.06 GHz (255 FSB) with the memory at 205 MHz, 2-3-2-5 and it booted and ran fine but crashed about half an hour later (I figured my RAM was being flaky at 205 MHz and tight settings, but I think this is not the case now). Dropped the CAS settings to 2.5-3-3-7 and still got flaky performance.

I got the Zalman to specifically help me with this chip, but it seems my 2.4B actually overclocks higher (it ran fine at 3.06 GHz and 3.1 GHz and was semi-stable at 3.2 GHz with the Zalman).

CPU temps are 28C on initial boot, 30-32C idle, 38-39C load (the Zalman seems to be doing it's job quite well!). How much voltage have people been putting into these new P4 C's? is 1.54V actual too low for going above 3 GHz or what? Or is my chip just a real loser?

 

[Duvie]

what bios??? I say it cause 250fsb limit and it sounds like a limiting factor other then your cpu cause of the "wall" effect you are stating, was quite common earlier with reports on abit Ic7 and Is7 mobos....is it a newer rev board with the new blue abit cooler and chipset...

I am buying one from a guy who had it running 270fsb at 1:1 with a 2.6c....

 

[Duvie]

Also 1.54v I thought was actually still below default of 1.55v for the p4c's....I have seen many have these up to 3.2-3.3ghz at default vcore...check out Thugsrook's new 2.4c on IC7....I think he got 3.4ghz at default vcore air cooled with a swiftech cooler.

One thing i would look at is voltages, and have you bumped up the vagp??? I hear the vagp sets the vcore for the northwood chipset. Bump it up to 1.6-1.7v and put the memory vdimm at 2.7v regardless...At these high fsb's everything seems to be drawing tons of power. The 2.4b did higher cause the fsb was lower and thus the chipset was not as oc'd.

 

[jiffylube1024]

Edit: double post.

 

[jiffylube1024]

Originally posted by: Duvie
Also 1.54v I thought was actually still below default of 1.55v for the p4c's....I have seen many have these up to 3.2-3.3ghz at default vcore...check out Thugsrook's new 2.4c on IC7....I think he got 3.4ghz at default vcore air cooled with a swiftech cooler.

One thing i would look at is voltages, and have you bumped up the vagp??? I hear the vagp sets the vcore for the northwood chipset. Bump it up to 1.6-1.7v and put the memory vdimm at 2.7v regardless...At these high fsb's everything seems to be drawing tons of power. The 2.4b did higher cause the fsb was lower and thus the chipset was not as oc'd.

Thanks for the replies Duvie. I think it is a slightly older IC7 with the older fan (it's a blue one and not too loud though) however it has been modded with the huge Zalman NB HS for better cooling.

I will check out what BIOS I'm running at.

I didn't realize the stock Vcore was 1.55V - in which case I'm still undervolting a bit. I guess I'll try putting 1.56-1.6V actual in there.

I have not tinkered with Vagp - I guess I'll bump that up. Hopefully it won't fry my Radeon 9800 or GF4 cards! Vdimm is at 2.65V - I'll up it to 2.7V.

Will post results later!

Edit: here's what SiSoft Sandra reports the board as:

MPS Version: 1.4
Version: 1.0
Bios version: ........ -17 (this is the newest one for this board I believe)

 

[THUGSROOK]

bump vdimm to 2.8v
disable CPC in bios

HTH

 

[Duvie]

Originally posted by: jiffylube1024

Originally posted by: Duvie
Also 1.54v I thought was actually still below default of 1.55v for the p4c's....I have seen many have these up to 3.2-3.3ghz at default vcore...check out Thugsrook's new 2.4c on IC7....I think he got 3.4ghz at default vcore air cooled with a swiftech cooler.

One thing i would look at is voltages, and have you bumped up the vagp??? I hear the vagp sets the vcore for the northwood chipset. Bump it up to 1.6-1.7v and put the memory vdimm at 2.7v regardless...At these high fsb's everything seems to be drawing tons of power. The 2.4b did higher cause the fsb was lower and thus the chipset was not as oc'd.

Thanks for the replies Duvie. I think it is a slightly older IC7 with the older fan (it's a blue one and not too loud though) however it has been modded with the huge Zalman NB HS for better cooling.

I will check out what BIOS I'm running at.

I didn't realize the stock Vcore was 1.55V - in which case I'm still undervolting a bit. I guess I'll try putting 1.56-1.6V actual in there.

I have not tinkered with Vagp - I guess I'll bump that up. Hopefully it won't fry my Radeon 9800 or GF4 cards! Vdimm is at 2.65V - I'll up it to 2.7V.

Will post results later!

I believe it is 1.55vcore as thugs stated unless he was reporting actual....Also the talk is the vagp at 1.6-1.7v should be as safe as any. If the northbridge gets its vcore off of the vagp then you can realize 50mhz over spec speed may require aboost like any cpu would....

Good luck!!!!

 

[jiffylube1024]

Originally posted by: THUGSROOK
bump vdimm to 2.8v
disable CPC in bios

HTH

Will do. What is CPC?

 

[THUGSROOK]

~command per clock

 

[jiffylube1024]

Originally posted by: THUGSROOK
~command per clock

OK I assume CPC is the command per clock (CPC) setting in the Advanced Chipset setting. I tried both enabled and disabled and it doesn't make a difference.

Discouraging results here:

260FSB (3.12 GHz) was attempted at the following voltages:

1.625V (1.56-1.58V actual) Failed before Windows
1.675V (~1.61V actual) Failed before Windows
1.725 (~1.66V actual) Failed before Windows

In all cases it seemed to go through the loading windows bar up until it's about to get right into windows, at which point it quickly BSOD/reboots every time.

With these results, I'm strongly inclined to believe this is a hard 250FSB limitation on my motherboard.

In this case, it would mean that my board, not my CPU is the limiting factor. And with my CPU temps, and the fact that I can run at 3.0Ghz at 1.47V actual, I strongly agree with this view. It must be my (earlier) revision IC7. Grr...

 

[Duvie]

This is more what I think I was talking about...

<Enhanced the system stability when over clocks the CPU FSB over 255 MHz with CPURAM = 1:1 setting. >

line 6 at Abit's aupport page for bios 17.... Bios version 17

i would mirror my settings to all that Thugs, minus the fsb set!!!

Hows the power supply???

have you tried the 3:2 mode and worst cas timings to make sure it isn't the memory???

 

[jiffylube1024]

Originally posted by: Duvie


i would mirror my settings to all that Thugs, minus the fsb set!!!

Done. Same results .

Hows the power supply???

430W Antec Trupower. Voltage rails are a tiny bit under for each but it's allowed me great overclocks in the past.

have you tried the 3:2 mode and worst cas timings to make sure it isn't the memory???

Yup.

 

[Duvie]

When was this mobo purchased originally??? I would sell it and pick up a new one...Like I said the guy I am buying from had his running 270fsb with 1:1 so though I may not get that high itis nice to know it has been there. Thug's is a rocket at booting into 291fsb....

 

[Duvie]

I don't know for sure but how are the swings of the vcore under load from idle??? I switched from my 430watt antec to 430 watt enermax and I was able to get to 3.33ghz stable when my antec would't go above 3.24ghz at same vcore....The rails were tighter as well....My 12v rail with my 2.4b @3.24ghz w/ 1.68v was teetering at 11.55-11.47 and vcore swings were .07v. I ran 2 hdds, 2 dvd/dvd-rw drives, AIW 8500DV, 2 pci cards, 2 usb devices.

I don't know if power is an issue in terms of the PSU or the mobo, but it is always something to think about....

 

[Duvie]

Originally posted by: jiffylube1024

Originally posted by: THUGSROOK
~command per clock

OK I assume CPC is the command per clock (CPC) setting in the Advanced Chipset setting. I tried both enabled and disabled and it doesn't make a difference.

Discouraging results here:

260FSB (3.12 GHz) was attempted at the following voltages:

1.625V (1.56-1.58V actual) Failed before Windows
1.675V (~1.61V actual) Failed before Windows
1.725 (~1.66V actual) Failed before Windows

In all cases it seemed to go through the loading windows bar up until it's about to get right into windows, at which point it quickly BSOD/reboots every time.

With these results, I'm strongly inclined to believe this is a hard 250FSB limitation on my motherboard.

In this case, it would mean that my board, not my CPU is the limiting factor. And with my CPU temps, and the fact that I can run at 3.0Ghz at 1.47V actual, I strongly agree with this view. It must be my (earlier) revision IC7. Grr...

Hmmm really!!!!

Here is the guy you bought it from and it states he ran it at 280fsb for 3 months!!!!



Abit IC7 Forsale Forum

 

[THUGSROOK]

i dont think its a board limit either.
my IC7 is old ~ and it does 290fsb with a 2.4C.
(bios v14)

try turning off more stuff.
~USB, onboard sound, etc....

 

[FPSguy]

Default voltage for a 2.4C is 1.525.

Did you try setting GAT to Auto-Normal-Auto-Disabled-Disabled?

If that doesn't fix it, I bet the problem is the memory. It seems that several ABit boards don't like Winbond CH-5 chips, and have problems with them using dividers other than 1:1 at FSBs over 255. You can read about it at forum.abit-usa.com.

 

[jiffylube1024]

Originally posted by: Duvie
Hmmm really!!!!

Here is the guy you bought it from and it states he ran it at 280fsb for 3 months!!!!



Abit IC7 Forsale Forum

That's him alright! Wow, I can't believe he ran it at 280FSB; I wonder what my problem is!!??

Even if I am getting 250FSB limitations, I don't begrudge him either way - it runs perfectly and 200 FSB (and well above) and I've been 100% pleased with the board.

Now, back to the issue. Apparently this board has been up to 280 FSB before (for 3 monts stable, no less).

Originally posted by: FPSguy
Default voltage for a 2.4C is 1.525.

Did you try setting GAT to Auto-Normal-Auto-Disabled-Disabled?

If that doesn't fix it, I bet the problem is the memory. It seems that several ABit boards don't like Winbond CH-5 chips, and have problems with them using dividers other than 1:1 at FSBs over 255. You can read about it at forum.abit-usa.com.

^ Wow. Nice find FPSguy! I have two sticks of 512MB PC3200 Winbond in here - one is BH5 and one is CH5. Looks like they could be the culprit!

I have a matched pair of 256MB PC3200 Kingmax TinyBGA here (I know it's not the greatest stuff) that I will try out in this board. And I can also try running just single channel with the BH5 also.

Looks like I have more work cut out for me, but it could be worth it. I'll post results later.

Thanks everyone!

On a side note, my Radeon 9800 np which I thought I fried is running better than ever. A word to the wise: BGA RAM chips are VERY touchy, and if you put pressure on the chips they will screw up your video quality (and crash).

 

[jiffylube1024]

Already in at 3.12 GHz! Time to see how far this chip will go now!

 

[jiffylube1024]

Edit: THIS IS WITH THE KINGMAX CHIPS IN THERE, crusing along past 280FSB (3/2 memory divider)

Boots at 3.36 GHz 1.675V in BIOS (1.62V actual - 1.60V during prime), runs Windows fine, Fails Prime instantly.

Boots at 3.24 GHz runs Windows fine, Fails prime instantly @ 1.600V (1.54V actual).

Boots at 3.24 Ghz, runs Prime for about 2 minutes @ 1.65V (1.6V actual).

CPU temps are low 30's.. Should I put more voltage in there (~1.7V, =~1.62V actual) ?

Good lord! Prime95 eats Voltage for breakfast, lunch, and dinner! At 1.65V in the BIOS, the actual reading in Windows (by MBM5) dips down to 1.58V! I think my 430W Antec Trupower is being pantsed by Prime. Checking the side of my Antec PSU (I think it's a slightly older revision) the rated voltage rails are as follows:

+3.3V 28A
+5V 36A
+12V 20A

No wonder I'm getting hammered so badly! Checking the Enermax website, their new 430W PSU's have 33A on the 12V line; even the 350W ones have 26A there!

I always buy Enermax but got this PSU in a bundle for a computer a year ago. I will definately be picking up a 430W Enermax tomorrow.

I'm currently flying through Prime @ 3.2 Ghz right now, 1.65V in the BIOS (gawd it just dipped to 1.57V).

With the new PSU I'm pretty sure I could make it work at 3.3 GHz with reasonable voltages. I wonder how high I will be able to post with that sucker...

 

[Duvie]

Originally posted by: FPSguy
Default voltage for a 2.4C is 1.525.

Did you try setting GAT to Auto-Normal-Auto-Disabled-Disabled?

If that doesn't fix it, I bet the problem is the memory. It seems that several ABit boards don't like Winbond CH-5 chips, and have problems with them using dividers other than 1:1 at FSBs over 255. You can read about it at forum.abit-usa.com.

That is confirmed!!! I believe the upper p4c like 3ghz and 3.2ghz are in fact 1.55v though....

nice find fpsguy!!

So are you running the kingmax chips???

 

[jiffylube1024]

updated my findings above, Duvie ^.

Boy, I've really learned a lot in the last day! I mean, I've been overclocking since my P3 700, but this is really good info. Now that the Winbond chips are a culprit for limiting my o/c, I have to make a decision. Try to trade for some comparable chips (are OCZ 3200 chips any good?) or just be happy with the Kingmax (which I doubt will keep me very long as they can't even run 200MHz! ).

Right now Prime is still chugging along, my Antec PSU is still allowing the voltage to sag down to 1.58V at 3.2 GHz and temps have climbed to a (still very low) 37C. Looks like the Zalman is turning out pretty well after all.

I can't wait to pop in the Enermax tomorrow. 65% more Amps to the P4's favourite rail should give me a lot more headroom.

 

[Duvie]

I see you are seeing what I saw with the PSU...

 

[jiffylube1024]

Originally posted by: Duvie
I see you are seeing what I saw with the PSU...

Yes, two year old Antec's suck!

 

[Duvie]

I see only .03v swings when I run my enermax....I set 1.61v and it never goes below 1.58v under prime95...Thugs and those guys kinda eyed the problem with 20a or less on the 12v rails and ocing over 3ghz. I think that is why he stepped up to the 550watt with well over 30amps.