Does the MB make " The Chip " or is it the CPU that makes it.

[Kenmitch]

I've been wondering about this for a couple of weeks.

I have an e5200 in a Gigabyte G41M-ES2L and am kinda paranoid to even mess with it as it has some very good results I'd say way above the norm when you take into account the vcore

I think she'd be a e5200s if intel made such a thing....This is what she'll do

2.00ghz (333x6) bios vcore .884 OCCT vcore .86 Temps 35*c idle 37* loaded
2.33ghz (333x7) bios vcore .900 OCCT vcore .88 Temps 35*c idle 38* loaded
2.66ghz (333x8) bios vcore .948 OCCT vcore .93 Temps 35*c idle 42* loaded

3.00ghz (333x9) bios vcore 1.028 OCCT vcore 1.01 Temps 35*c idle 44* loaded
3.33ghz (333x10) bios vcore 1.108 OCCT vcore 1.07 Temps 35*c idle 50* loaded
3.66ghz (333x11) bios vcore 1.188 OCCT vcore 1.15 Temps 35*c idle 60* loaded
4.00ghz (333x12) bios vcore 1.300 OCCT vcore 1.27 Temps 35*c idle 72* loaded

4.16ghz (333x12.5) bios vcore 1.380 OCCT vcore 1.34 Temps 38*c idle 84* loaded

I'm wondering if the chip would most likely have the same results on any board that is capable of overclocking.

Any input on past OC adventures would be of interet to me.

Thanks,
Ken

 

[alyarb]

not sure if this is a brag thread or what, but if you think the G41 board won't allow you to achieve the same ELVs, you're probably wrong and it's worth a shot. you know you don't have to stress the FSB and you're just ticking the multi down, so what are you waiting for?

oh, right, the brag thread.

 

[Lemon law]

From what I have read, its the cpu and not the board. Although there may be some board cpu interactions.

Because the amount of leakage of voltages between cpu layers is impossible to totally control in the manufacturing process. But then again, it would apply to all the other IC's in the mobo also, but the smaller the namometer fabs of the IC, the more that may apply. Basically one is growing crystal, and crystals always have random flaws or holes where it does not crystallize perfectly.

But that would be more opinion and guesses and not proven fact on my part.

But you could test it statistically, but you would have to have many motherboards and cpu's to even approach a 95% confidence level.

 

[Yukmouth]

Unless you're a fan of horrendously cheap motherboards, CPU all the way.

 

[Zap]

IMO it is the CPU... motherboard willing... as long as the board doesn't get retarded about memory multipliers with low FSB chips.

My server box is running an E5200 at 1.2GHz and 0.788v (detected in BIOS, set to 0.8v). Seems like these chips have quite the operational voltage range.

 

[Kenmitch]

Brag thread? Whatever

I guess what I meant was do you guys find that the CPU will have similar results from MB to motherboard.

Or that it varies from MB to MB alot.

I kinda wanna do some shifting around with some chips for testing purposes but thought I'd ask the
question hoping others would kick in some past experiences on the subject.

Thanks

 

[alyarb]

what i was saying is 333 FSB isn't going to put any motherboard to the test. undervolting doesn't stress anything either.

 

[Kenmitch]

So I guess most likely this chip would do similar in another MB that allowed voltage and fsb adjustments as it likes the 1333fsb.

 

[Yukmouth]

Originally posted by: Kenmitch
So I guess most likely this chip would do similar in another MB that allowed voltage and fsb adjustments as it likes the 1333fsb.

Yes, though voltages may vary.

 

[Shmee]

you must have a pretty good chip IMO. Now, boards with good LLC may allow you to lose a few milivolts, but I would say at this point you are more CPU/Temp limited. I think youll still be able to go further, unless a massive vcore boost is needed. Well, actually, probably not, considering 84 C is pretty high. May want a better cooler to go farther.

As for past OC stories, me and my friend faxon paired one of the first E5200's with an xfx 680i LT, not the best oc board. Most we could get was 3.3 stable with about 1.4 V. Same results on an awesome new UD3P. In short, sometimes you will be chip limited, sometimes board limited, sometimes temp/cooler limited. I would say here you are lucky to have a good chip, and the board isnt too shabby, and even if it was, you have a 12.5 multi so dont worry too much. Just need better cooling for temps under load.

However, if you were ocing a Q9550, you would certainly be board limited. With a stock FSB of 1333 and an 8.5 multi, plus additional north bridge stress, you would need a good board for ocing a Q9550 to 4 or more. For cheap, high multi dualcores, it is much less of an issue.

 

[Kenmitch]

Originally posted by: Shmee
you must have a pretty good chip IMO. Now, boards with good LLC may allow you to lose a few milivolts, but I would say at this point you are more CPU/Temp limited. I think youll still be able to go further, unless a massive vcore boost is needed. Well, actually, probably not, considering 84 C is pretty high. May want a better cooler to go farther.

As for past OC stories, me and my friend faxon paired one of the first E5200's with an xfx 680i LT, not the best oc board. Most we could get was 3.3 stable with about 1.4 V. Same results on an awesome new UD3P. In short, sometimes you will be chip limited, sometimes board limited, sometimes temp/cooler limited. I would say here you are lucky to have a good chip, and the board isnt too shabby, and even if it was, you have a 12.5 multi so dont worry too much. Just need better cooling for temps under load.

However, if you were ocing a Q9550, you would certainly be board limited. With a stock FSB of 1333 and an 8.5 multi, plus additional north bridge stress, you would need a good board for ocing a Q9550 to 4 or more. For cheap, high multi dualcores, it is much less of an issue.

Thanks for the input

In the current board the highest she'll load windows is 4.25ghz @ 1.364 didn't do any stress testing tho as the cooler crapped out at 4.00ghz. Pretty sure the chip woulld fly under 1.400 with a better cooler.

I can see the point of the high end board needed for the Q9550 or any other 1333fsb chips. Guess maybe it's intels way of cutting down on the overclocking high fsb low multiplier.

I've got a couple more e5200's to check out on Tuesday. I'm thinking about doing a BSEL mod to run a chip on an intel board. I don't want to waste this chip as it just wouldn't make the 1333fsb without a voltage boost on the intel board. And dropping to 1066fsb the chip would be getting way more voltage than needed.

Guess as silly as it sounds I'm hoping one of my other e5200's will do about 3.5ghz at stock voltage making it a better candidate for the 1066fsb BSEL mod which would be 3.33ghz in a non overclockable MB.

Thanks

 

[heyheybooboo]

Topic Title: Does the MB make " The Chip " or is it the CPU that makes it.

Yes.

 

[firewolfsm]

With the memory controller on the motherboard, it makes a difference at high FSBs, but 333 is standard, the rest is your cpu.

 

[aigomorla]

Id say about 80% chip
20% board...

 

[Kenmitch]

This is what I learned so far.

The chip is born " The Chip " just gotta let her out of the cage that Intel or AMD put her in

If the chip will make the jump from 800fsb -> 1333fsb got it made depending on ram and dividers due to no stressing of the NB and SB chips.

A high quality board is required for the Quad cores due to the need of very high fsb which puts alot of stess on the MB.

If the system reboots when stress testing trying to relax the timings on your overclocked ram may help....Found this one on my own testing another e5200 that made the jump to 1333fsb. Damn ram dividers don't like to play well with DDR2-800 mem.

It is a possibility that the cooler may wind up costing more or very close to the price of the chip depending on the temps and vcore required.

The above example of the e5200 would be an above average chip.

While whipping up another e5200 for testing today I thought I'd see what the absolute highest stable overclock would be on this chip....Very cool today with a way lower ambient temp so my cooler seemed to drop load temps about 10*c

Above 4.0ghz I only had to disable the EIST to keep her humming along. With a vcore adjust that is. At least I didn't have to disable all speed step features in bios.

The chip is currently running at 4.25ghz bios vcore 1.396 OCCT vcore 1.36 temp hovers around 80*c in Intel Burn Test and around mid 60's in OCCT.

Guess the theoretical max with this chip is 4.28ghz on this board....FSB wall above 343fsb setting. Didn't want to feed more vcore or mess with other voltages to try any higher.

Chip is going back to 3.33ghz (333x10) bios vcore 1.108 OCCT vcore 1.07 Temps 35*c idle 50* loaded once OCCT is done


Thanks

 

[Lemon law]

I can not answer in the affirmative to the Kenmitch conjecture of " It is a possibility that the cooler may wind up costing more or very close to the price of the chip depending on the temps and vcore required."

The point being, the intel stock heat sink and fan that comes with the E5200 is decidedly inferior to the hear sinks intel furnishes for the pentium D's. As the latter has a copper core and a larger mass and area of fins.

And I know for a fact that you can purchase better than intel pentium D heat sinks and fans for $20.00 RETAIL. And something like a $30.00 arctic cooler pro would positively be probably be overkill for an E5200. But point taken, the $80.00 nuke the world heat sink and fan option exists for the E5200.

Sad to say, that $20.00 retail better than intel stock pentium D is barely adequate to keep a 95+ watt intel pressler core 915 temperatures from soaring out of control at stock speeds, but would probably drop E5200 overclocked load temps at least 5+C
compared to its its stock hear sink and fan. And in terms of the intel published red line, its 63.3C for the pressler and 74.1C for the
E5200.

 

[Kenmitch]

Originally posted by: Lemon law
I can not answer in the affirmative to the Kenmitch conjecture of " It is a possibility that the cooler may wind up costing more or very close to the price of the chip depending on the temps and vcore required."

The point being, the intel stock heat sink and fan that comes with the E5200 is decidedly inferior to the hear sinks intel furnishes for the pentium D's. As the latter has a copper core and a larger mass and area of fins.

And I know for a fact that you can purchase better than intel pentium D heat sinks and fans for $20.00 RETAIL. And something like a $30.00 arctic cooler pro would positively be probably be overkill for an E5200. But point taken, the $80.00 nuke the world heat sink and fan option exists for the E5200.

Sad to say, that $20.00 retail better than intel stock pentium D is barely adequate to keep a 95+ watt intel pressler core 915 temperatures from soaring out of control at stock speeds, but would probably drop E5200 overclocked load temps at least 5+C
compared to its its stock hear sink and fan. And in terms of the intel published red line, its 63.3C for the pressler and 74.1C for the
E5200.

All of my testing was done with a Cooler Master Vortex 752 ($25.00 at fry's) which does ok to a point. I bought it originaly to put in my HTPC system as it was the tallest that would fit in it. All the temps are from Intel Burn Test with an ambient room temp of 78*f at that time.

Looks like the current chip I'm testing is gonna be another 4.00ghz chip although with a little higher vcore....At least on the G31M-ES2L board that I purchased with intent to make a Hacintoxh system out of it.

 

[faxon]

Originally posted by: Kenmitch

Originally posted by: Shmee
you must have a pretty good chip IMO. Now, boards with good LLC may allow you to lose a few milivolts, but I would say at this point you are more CPU/Temp limited. I think youll still be able to go further, unless a massive vcore boost is needed. Well, actually, probably not, considering 84 C is pretty high. May want a better cooler to go farther.

As for past OC stories, me and my friend faxon paired one of the first E5200's with an xfx 680i LT, not the best oc board. Most we could get was 3.3 stable with about 1.4 V. Same results on an awesome new UD3P. In short, sometimes you will be chip limited, sometimes board limited, sometimes temp/cooler limited. I would say here you are lucky to have a good chip, and the board isnt too shabby, and even if it was, you have a 12.5 multi so dont worry too much. Just need better cooling for temps under load.

However, if you were ocing a Q9550, you would certainly be board limited. With a stock FSB of 1333 and an 8.5 multi, plus additional north bridge stress, you would need a good board for ocing a Q9550 to 4 or more. For cheap, high multi dualcores, it is much less of an issue.

Thanks for the input

In the current board the highest she'll load windows is 4.25ghz @ 1.364 didn't do any stress testing tho as the cooler crapped out at 4.00ghz. Pretty sure the chip woulld fly under 1.400 with a better cooler.

I can see the point of the high end board needed for the Q9550 or any other 1333fsb chips. Guess maybe it's intels way of cutting down on the overclocking high fsb low multiplier.

I've got a couple more e5200's to check out on Tuesday. I'm thinking about doing a BSEL mod to run a chip on an intel board. I don't want to waste this chip as it just wouldn't make the 1333fsb without a voltage boost on the intel board. And dropping to 1066fsb the chip would be getting way more voltage than needed.

Guess as silly as it sounds I'm hoping one of my other e5200's will do about 3.5ghz at stock voltage making it a better candidate for the 1066fsb BSEL mod which would be 3.33ghz in a non overclockable MB.

Thanks

yea he wasnt quite accurate on that 100%. i was able to get it stable on the 680i using an XFX bios with higher CPU voltage, but some other things didnt quite work right. we flashed the board to an EVGA bios, and while we werent able to get it stable past 3.2 with 1.392vcore (CPUz) it was easier to get fine tuned than with the XFX bios due to finicky memory settings that went away with the EVGA bios. on the UD3P this wasnt an issue, and i was able to get 100MHz more out of the CPU with the same Vcore. then there's my Q9650, in which case it is very much a board limiting setup, where i was having to tweak board settings more than cpu vcore to get it that last 200MHz to 4.2. 4GHz was easymode with 1.296vcore, but i had to change the memory divider, increase the NB FSB and VTT, as well as increasing the vcore to 1.376 to get it up to 4.2GHz. with a lesser board this may not have been possible for 24/7 stability

 

[filibusterman]

Originally posted by: Kenmitch
This is what I learned so far.

The chip is born " The Chip " just gotta let her out of the cage that Intel or AMD put her in

If the chip will make the jump from 800fsb -> 1333fsb got it made depending on ram and dividers due to no stressing of the NB and SB chips.

A high quality board is required for the Quad cores due to the need of very high fsb which puts alot of stess on the MB.

If the system reboots when stress testing trying to relax the timings on your overclocked ram may help....Found this one on my own testing another e5200 that made the jump to 1333fsb. Damn ram dividers don't like to play well with DDR2-800 mem.

It is a possibility that the cooler may wind up costing more or very close to the price of the chip depending on the temps and vcore required.

The above example of the e5200 would be an above average chip.

While whipping up another e5200 for testing today I thought I'd see what the absolute highest stable overclock would be on this chip....Very cool today with a way lower ambient temp so my cooler seemed to drop load temps about 10*c

Above 4.0ghz I only had to disable the EIST to keep her humming along. With a vcore adjust that is. At least I didn't have to disable all speed step features in bios.

The chip is currently running at 4.25ghz bios vcore 1.396 OCCT vcore 1.36 temp hovers around 80*c in Intel Burn Test and around mid 60's in OCCT.

Guess the theoretical max with this chip is 4.28ghz on this board....FSB wall above 343fsb setting. Didn't want to feed more vcore or mess with other voltages to try any higher.

Chip is going back to 3.33ghz (333x10) bios vcore 1.108 OCCT vcore 1.07 Temps 35*c idle 50* loaded once OCCT is done


Thanks

I usually just disable all speedstep and EIST and such for overclocking. Is there any real reason to leave them on if the rig is going to be running all the time for say distributed computing or are those features more for conserving power in times if idle? Also, I guess I have not explored enough on my overclocks to tell what temperature levels are acceptable for what components but I have never seen a cpu temp> low 60C. But I guess I am more concerned with the NB and SB temps and even more concerned with the sensor accuracy on those chips.

 

[Kenmitch]

Originally posted by: filibusterman

I usually just disable all speedstep and EIST and such for overclocking. Is there any real reason to leave them on if the rig is going to be running all the time for say distributed computing or are those features more for conserving power in times if idle? Also, I guess I have not explored enough on my overclocks to tell what temperature levels are acceptable for what components but I have never seen a cpu temp> low 60C. But I guess I am more concerned with the NB and SB temps and even more concerned with the sensor accuracy on those chips.[/quote]

I always skip the disable all speedstep items in overclocking guides as I'd prefer to get the max overclock possible with the speedstep enabled. Maybe I have the illusion that it would save energy at the idle times or maybe prolong the chips life.

The above MB and e5200 combo was purchased to replace the wifes websurfing email machine so not much power was required and stock speed would still be overkill.

I was bored one day and decided to see what the chip would do. Just thought I'd see if I could get some more bang for the buck as they say. I wasn't expecting such a huge overclock out of the chip. Figure maybe around 3.5 ghz max.

Guess now I'm not sure where this chip will wind up as it seems like a waste of a great chip in her machine.

Thanks