Core2Duo = Core2Quad?

V

Vorret

[H]ard|Gawd
Joined
Aug 26, 2003
Messages
1,100
If a motherboard is able to run a Core2Duo , can we safely assume it can run a Core2Quad?

Just wondering, I want to cheaply upprade my GF cpu without having to change the motherboard and ram.

Thanks
 
No.

There are power differences, and termination differences in some of the high speed signaling and on top of that there are different generations.

Just because your motherboard can power a 65nm Conroe based C2D, doesn't mean it can run a 45nm Penryn based C2Q. Some cases they can, IIRC my ASUS p5B can with a bios update, and I think some gigabyte boards ,but that isn't the case with all manufacturers.

Why not post the exact mobo model, and exact cpu models you are working with here...?
 
Vorret said:
If a motherboard is able to run a Core2Duo , can we safely assume it can run a Core2Quad?

Just wondering, I want to cheaply upprade my GF cpu without having to change the motherboard and ram.

Thanks
Click to expand...

Whether it can RUN a core 2 quad or not, or what variation of the quad (remember there are Kentsfield and Yorkfield quads) depends purely on the BIOS support.

Overclocking them is a different story.

If your board has a 4 pin CPU 12v connector, you do NOT want to overclock quads on high voltage, or you risk burning out the connector. You can overclock them to around 4 ghz on default vcore (if they will run that high) without crashing...but if you want to pump the voltage, you *need* an 8 pin connector. Boards with 8 pin cpu connectors were designed for overclocking quads.

You also have to worry about boards getting low FSB with locked quads, due to them missing two GTL ref signals on the board, which again means that the board was made for duals. But that only applies to very old boards, e.g. 965 or 975 chipsets. X38 and P45 and so on clock quads fine.
 
Wow, more complicated than I expected.
I'll look into the model number when I get home tonight, I know it's a gigabyte and I remember something about -DS3 at the end but that's pretty much it :)

Thanks for the info!
 
Vorret said:
Wow, more complicated than I expected.
I'll look into the model number when I get home tonight, I know it's a gigabyte and I remember something about -DS3 at the end but that's pretty much it :)

Thanks for the info!
Click to expand...
It's not all that bad, or maybe that's just because I was heavy into doing lots of upgrades for people n that time period and familiar with the time-line of products there.

GA-945P-... most likely not
GA-965P- possibly with bios update
GA-P35- almost definitely
GA-P45- definitely
 
Menelmarar said:
It's not all that bad, or maybe that's just because I was heavy into doing lots of upgrades for people n that time period and familiar with the time-line of products there.

GA-945P-... most likely not
GA-965P- possibly with bios update
GA-P35- almost definitely
GA-P45- definitely
Click to expand...

P35 can for certain, 965P is limited to 65nm quads.
 
MrMike said:
P35 can for certain, 965P is limited to 65nm quads.
Click to expand...
965P chipset is not limited to 65nm only. It depends on the motherboard design and manufacturer support.

Certain model/revisions of ASUS, Gigabyte, and other quality motherboards with bios update can run the 45nm.
 
Menelmarar said:
965P chipset is not limited to 65nm only. It depends on the motherboard design and manufacturer support.

Certain model/revisions of ASUS, Gigabyte, and other quality motherboards with bios update can run the 45nm.
Click to expand...

I know.

We were specifically talking about the GA-965P-DS3.
 
LOL, I just checked and there's a few dozen "DS3" LGA 775 board models on gigabyte's site.

As mentioned above, it really depends on the chipset. Some are not compatible with quads, while others are limited to older quads, or lower power quads (95w).
 
Falkentyne said:
If your board has a 4 pin CPU 12v connector, you do NOT want to overclock quads on high voltage, or you risk burning out the connector. You can overclock them to around 4 ghz on default vcore (if they will run that high) without crashing...but if you want to pump the voltage, you *need* an 8 pin connector. Boards with 8 pin cpu connectors were designed for overclocking quads.
Click to expand...

This is not accurate. I've overclocked/overvolted MANY (dozens) quads with boards using a 4-pin connector and I've had a problem with exactly 0 of them.
 
RamonGTP said:
This is not accurate. I've overclocked/overvolted MANY (dozens) quads with boards using a 4-pin connector and I've had a problem with exactly 0 of them.
Click to expand...
This.

While I can't attest to overclock/overcolt capability.

You won't "risk burning out the connector" the current difference just isn't there, the resistance may rise, and you may get some voltage droop that you won't be able to overcome, and thus limit your overclock. But you won't burn anything out.

If you burn out that connector, something else catastrophic happened.
 
DS3 was crap from my experience. Wouldn't work with a bunch of processors. Eventually just died.
 
Menelmarar said:
This.

While I can't attest to overclock/overcolt capability.

You won't "risk burning out the connector" the current difference just isn't there, the resistance may rise, and you may get some voltage droop that you won't be able to overcome, and thus limit your overclock. But you won't burn anything out.

If you burn out that connector, something else catastrophic happened.
Click to expand...

If you do a search, quite a few people have burnt out the 4 pin by overclocking quads at high volts over time. There's a reason why boards that were designed specificially for quads and have high phase regulation have 8 pin connectors. The more connectors you have, the less heat goes through each pin. You do know this, right? This is basic electrical knowledge. For example, drawing 200W through one wire will generate twice as much heat ON that wire as drawing 200W through TWO wires, because the current isn't being spread out.

Just because it didn't happen to YOU does NOT mean that it might not happen to someone else. This is like saying that just because you put 1.5v into a 45nm C2Q and experienced no degradation that it's completely safe for everyone else, which is completely wrong--not everyone will have the same results you have. EVEN if the exact same hardware is being used. This same "hey it's safe for me, it will be safe for you" type of attitude cost me several Pentium 4 chips back in the day, before SNDS/GNDS knowledge was available that people started seeing degradation even at vcore as low as 1.6v. (and everyone was busy saying that as long as you stay at 1.7v or lower, you were safe....)

I burnt out the 4 pin on my PSU and my P5WDH by doing precisely that. And no, it was NOT a catastrophic occurance. I had been running at 4 ghz and 1.4v (vdroop modded) for a long time (originally started at 1.36 three years ago) and the huge power draw simply heated up the connector too much. Eventually it scorched slightly from all the heat. When I tried 4.2 ghz at 1.45v with a new QX recently, that was just too much. (the connector became completely burnt in a few days). Before the 4.2, the slow scorching was a gradual process over time, not something that happened suddenly. Using a board not with official support for quads, and running very high vcore for that cpu.

I know it was the power draw that caused it, because my PSU's fan would ramp up noticably when running prime or linpack.

And furthermore, remember that this was a QX9650 C0 stepping. Those drew LOTS Of power and ran quite hot.

Remember the P5WDH was absolutely *NOT* designed for 45nm quads. At ALL. It could run them with a BIOS update, but there were various issues: Speedstep and EIST/C1 didn't work, half multipliers didn't work, and FSB was lower than on many other boards, because it only had GTL lanes for 2 cpus, not for four, so the 2 requested lanes had to get their power from somewhere else (Freecableguy actually wrote an article explaining this, and showing the results of his testing).
I'm using the same PSU now on my SB, but using the 8 pin connector, and no problems at all.

So, who knows? Maybe it was the motherboard to blame (which didn't support 45nm fully). Maybe the PSU (doubtful, as it was a PC P&C 750W, still in service on my SB). Maybe it was a shift in the magnetic pole, combined with a realignment of some radio antennae at a local airport...

So anyway, please don't try to spread your information as gospel just because you had no problem. Perhaps boards that are 45nm ready and that have all four GTL lanes and have better voltage regulation won't have issues. But the OP was clearly asking about boards that might NOT have official quad support. Remember, the old P5wDH didn't even work with core 2 duals when it first came out..it needed a BIOS update for those...and the older revisions (1.2 and older) wouldn't work properly with ANY 45nm CPU...vcore range would start at 1.4, only auto gave correct vcore...
 
Last edited:
By that same token, just because it happened to someone else, doesn't mean it will happen to you. The fact of the matter is that the odds of something like that happening are VERY slim so the claim that you "need" an 8-pin connector is not an accurate statement. People have also burned their CPU's by overclocking, does that mean you NEED to keep it at default speeds?
 
I can accept your counter argument, but I think it's better to be safe than sorry.
But I can agree that using a board that fully supports 45nm quads, and a 4 pin connector, is alot safer than using a board like a board with only non official "YMMV" support.
(there were so many issues with p5wdh and 45nm, the XS thread was littered with problems unless you had a 1.3 or 1.4 version).

I was just trying to tell him that if he finds his board can accept a quad, even though it's a "dual" only board (quads possibly supported later in its life cycle), to be careful with overclocking.

Anyway, all of the boards that natively support 45nm and are NOT budget line boards have 8 pin power connectors, not 4-pin. They can USE 4 pin connetors (the higher end P5Q boards even have 4 pins covered to show people where to plug in the 4 pin connector), but it's common knowledge that you're supposed to use an 8 pin for overclocking. That's one reason why it's there :)

I still feel that my post was hitting the mark, since the OP was asking about C2D boards running quads, that may not be P35 or newer, and that was pretty much the same situation as me running a QX9650 on a board that wasn't designed for it.

I know that cheap P45 boards have 4 pin connectors (like the DS3L), but at least they natively support quads. But you're still running a risk using a 4 pin connector and throwing vcore at the chip.

So I was saying, if he finds his C2D board did support quads with a bios update, but was not a chipset that was natively 45nm quad ready (965, 975X), to be careful about pushing the current too high with a 4 pin connector.

Remember, mine was fine for over two years; going past 1.375v at 4 ghz with vdroop removed was basically the turning point...
 
For example, drawing 200W through one wire will generate twice as much heat ON that wire as drawing 200W through TWO wires, because the current isn't being spread out.
Click to expand...
Assuming all wires/pins are identical halving the number of wires/pins will mean FOUR TIMES as much heat on each wire/pin. Remember that for a given resistance power is proportional to the sqare of current.

Now as for getting into some actual sums the contacts in the four pin connector are specified as molex "44476-1112" or equivilent. According to molex those have a rating of 11A so with two 12V pins that is 264W. Current rating for the wire will depend on the wire type and how it's installed but afaict it will generally be higher than the rating of the connector. Based on that it should handle a reasonable ammount of overclocking but extreme overclocking could probablly push it over.

However those current ratings are probablly based on a cool environment (molex don't seem to say what temperature they were based on but toom temperature is a reasonable assumption) based on high quality connectors. If anything is even slightly marginal it doesn't take much to push a connector from "heavilly loaded but OK" to "gradually deteriorating from heat"
 
Ouch....then I stand VISIBLY corrected.
And thank you. But this simply shows why it's much better, and saner, to use an 8 pin connector than a 4 pin.

Now the question is, just how much power is a QX9650 @ 4.2 ghz and 1.45v using?
I can't find any hard numbers, but if stock speed fully loaded is 260W with prime95, running it at 4.2 ghz and 1.45v could easily be over 350W, far more than the 4 pin connector was designed for. Which explains why, after 3 years of abuse (4 ghz and 1.3625 to 1.4v 24/7, later on), 1.45v was good game.

*edit*: This formula OC Wattage = TDP * ( OC MHz / Stock MHz) * ( OC Vcore / Stock Vcore )^2
doesn't tell me anything at all. Plugging in values results in 244 W, assuming 130W TDP, and 1.45v and 4.2 ghz but other reviews had the chip using 260W at stock vcore, stock speeds when running prime95 full load...so I have no idea what's going on there.

heres another chart showing consumption, at stock settings: http://www.techspot.com/review/75-intel-core2-extreme-qx9650/page6.html 214W load at stock; they didn't say what they used though. Definitely not Linpack...
 
Last edited:
I'm pretty positive that's total system power consumption, not just the CPU.
 
214W system power consumption? (going from the last link) Definitely not :)
Unless a videocard and hard and optical drives were all capable of using less than 100W together
Even a video card at 2d settings can't combine with drives for that low :D Maybe if its a 5200 or 4200 card...

Rather sure its the CPU.

BTW just reading a thread about "SB dying/voltages" caused me to run into another post about burnt 4 pin connectors, and I wasn't even looking for it....

Ben_Lamb
itanic Join Date: Oct 2009
Posts: 203

--------------------------------------------------------------------------------

Funny how Intel is recomending the same 1.375max volts as the 45nm processors, the 32nm process must be quite durable. Then again they used to recommend 1.5vmax for 65nm chips like the q6600 and that was not healthy for chip or motherboard, burnt the cpu power socket out on mine (4pin). I think anything over 1.4v is definately no good for 24/7 but with a three year warranty who cares ? If Intel get enough returned K series chips overclocking could be phased out or they will have to put a voltage limiting circuit on chip !

-----
http://www.semiaccurate.com/forums/showthread.php?t=4064&page=3

So yeah its not as rare as you may think, especially on boards that were made for 65nm (such as mine, which burnt slowly with 45nm high clocks/vcore)
 
Last edited:
Falkentyne said:
214W system power consumption? (going from the last link) Definitely not :)
Unless a videocard and hard and optical drives were all capable of using less than 100W together
Even a video card at 2d settings can't combine with drives for that low :D Maybe if its a 5200 or 4200 card...

Rather sure its the CPU.

BTW just reading a thread about "SB dying/voltages" caused me to run into another post about burnt 4 pin connectors, and I wasn't even looking for it....

Ben_Lamb
itanic Join Date: Oct 2009
Posts: 203

--------------------------------------------------------------------------------

Funny how Intel is recomending the same 1.375max volts as the 45nm processors, the 32nm process must be quite durable. Then again they used to recommend 1.5vmax for 65nm chips like the q6600 and that was not healthy for chip or motherboard, burnt the cpu power socket out on mine (4pin). I think anything over 1.4v is definately no good for 24/7 but with a three year warranty who cares ? If Intel get enough returned K series chips overclocking could be phased out or they will have to put a voltage limiting circuit on chip !

-----
http://www.semiaccurate.com/forums/showthread.php?t=4064&page=3

So yeah its not as rare as you may think, especially on boards that were made for 65nm (such as mine, which burnt slowly with 45nm high clocks/vcore)
Click to expand...
Yes that's full system power.

That link to SA is hysterically laughable... Really?... REALLY!?.... LOL
Do you know how many Q6600's shipped and are still running today without this problem? With Intel's rep for quality, and this guy's obvious IQ, it's most likely user error as such as anything.
 
You must log in or register to reply here.
Back
Top