Quad Core with Different Multipliers

[xnem3s1sx]

I just got a Core 2 Quad Q9400, to replace my old q6600, (swapped with a friend and paid the difference) and the first thing I noticed was that, although the temps were lower (CoreTemp) the third core was stuck at 35 degrees. It went up if the other cores went above 35, but never below. I wasn't really concerned about it, until I went to Everest Ultimate 4.60 and opened the CPUID. It tells me that the first 2 cores are at 2.66 GHz, and 8x multiplier, but the second 2 are on a 6x Multiplier. I went to the BIOS and made sure that it was set at 8x, it was, so I changed it from auto to manual, with the same settings. But I'm still reading a 6x multiplier.
So I have to ask, is this even possible? I know that the C2Q's are really 2 dual cores on one chip, but I thought that the multiplier applied to all the cores. Is there anything I can do to test this?

 

[blackwidow_rsa]

Try cpu-z
www.cpuid.com

 

[xnem3s1sx]

I did, but CPU-Z doesn't allow you to see the individual stats of each CPU.

 

[soark]

Yes it does..click with the right button of the mouse anywhere in the CPU Tab and you can choose the core to view the information. You can also launch several CPU-Z and use one for each core.

Hope this helps but i have a Q6600 and i don't believe that the multipliers can be different among the cores.. the temperature thing is the same that exists in a large number of 45nm quad cores,cant really do anything I think, guide the temps from the other cores

GL

 

[V3NOM]

I don't believe it. Everest is wrong.

 

[bobwya]

If you have Speedstep 2 enabled in the BIOS then it is most likely that one (dual-core) die is being dropped down to the lowest possible multiplier (6x) with your system processes shunted to the other die. If you disable (all!!) the obscure speedstep (4 letter acronyms) settings in the BIOS your cores should stay at the 8x multiplier.

Bob

 

[xnem3s1sx]

Used CPU-z and it says the same thing, as for the speed step, I thought so too, but It never goes to 8, regardless of what program I am running.

 

[xnem3s1sx]

And I've tried disabling all of that, but to no avail.

 

[bobwya]

Used CPU-z and it says the same thing, as for the speed step, I thought so too, but It never goes to 8, regardless of what program I am running.

Even if you manually assign a CPU bound process to one of those 2 "idle" cores?

Bob

 

[xnem3s1sx]

correct, i have run folding at home and prime 95 on those cores, but they still stay at 6x

 

[bobwya]

OK thats fair enough.

■ Did you clear the BIOS after installing the new CPU?
■ Have you an updated your MB BIOS to the newest version?
■ Is the matter a software bug in CPUz (e.g. reading the clocks on dual-die Wolfie)?
Does a CPU bound process that will run in the CPU cache actually take longer on the "supposedly" lower clocked cores?? (Thinking SuperPI here perhaps)

Bob

 

[xnem3s1sx]

Yes, i have updated the BIOS. What do you mean by clear though? I reset to the optimized (I have an MSI P7N Platinum) And i thought about testing whether or not it actually takes any longer, but what program would I use to check that, it would need to be single core.

 

[bobwya]

Yes, i have updated the BIOS. What do you mean by clear though? I reset to the optimized (I have an MSI P7N Platinum) And i thought about testing whether or not it actually takes any longer, but what program would I use to check that, it would need to be single core.

You clear the BIOS by taking out the 3V round lithium battery mounted on the MB and turning off the PSU. Then replace the battery and switch on. You can also do this by farting around with CMOS clear jumpers but this is a bit of a pain in the ass...
Some motherboards have a clear CMOS button which is handy (I think you press them with the PSU off to interrupt the power connection from the CMOS backup battery).

I suggested a program already: SuperPI (it is single threaded). Just run it with a small data-set and then run it again. Take the second result (so that it will be fully contained within the CPU cache). Run this on each core one at a time with background **** turned off. Process Explorer is handy for suspending or kill off "deadwood" processes vs the crap builtin task manager:
http://technet.microsoft.com/en-us/sysinternals/bb896653.aspx

Bob

 

[xnem3s1sx]

Ok, I cleared the BIOS, and tested the actual speed, but the 2 cores are still at 6x. My guess is its probably just a bad proc. I'm gonna email Intel, but if anyone knows something else, it would be much appreciated.

 

[cjl]

E-mailing Intel sounds like the best way to go.

As for everest being wrong? It might not be in this case, but it reads my i7 wrong all the time.

 

[jas69]

Did you find the problem xNEM3S1Sx.
I have the same problem with my Q9550. I've got 1 core @ 8.5 multiplier, 1 @ 8.0 & 2 @ 6.0 as cpuz is showing. Thats with everything disabled. Even under 3dmark cpu test, the multipliers wont scale up.

 

[zforce67]

Did you find the problem xNEM3S1Sx.
I have the same problem with my Q9550. I've got 1 core @ 8.5 multiplier, 1 @ 8.0 & 2 @ 6.0 as cpuz is showing. Thats with everything disabled. Even under 3dmark cpu test, the multipliers wont scale up.

I too am having the same problem with my q9550. 2 cores at 8.5 whilst the other 2 cores are at 6 with a steady temp of 43c(barely ever goes up or down).

I've tried everything.

Anything?

 

[someguy7]

Screenshot of cpuz at idle and under load. 2 people with same problem, no pictures, no follow up from the 1st person.

 

[bobwya]

+1 It's like these guys don't actually want any help!!

Has any of you guys solved the problem... Even if this included an RMA replacement!! That could indicate a fault in a batch of Q9550 CPUs perhaps??!!

Bob

 

[zforce67]

Hi,

Here is a screenshot of the first(and second core:


The next is the problem screenshot with the 3rd(and 4th core):






My board is an Asus P5W DH Deluxe, originaly configured for 1066/800 FSB, the new bios are suppose to support 1333fsb in an overclocked manner. Without any tinkering on my part the board automatically configures/overclocks to the CPU's original cinfig of 2.83/1333, yet even then the 3rd and 4th core/s multipliers are set to 6.

Hope this helps.

(Corrected Photos)

 

[The_Blood_Raven]

The reading is false, there is no way the motherboard could handle 530+ Mhz of FSB! Download 3dmark Vantage and compare your CPU speed results online.

 

[zforce67]

Indeed, it doesn't make sense, the most the board would theoritically allow for the FSB is 500MHz. But the thing that gets me is the fact that the two cores(3/4) temperatures are stuck at 43c(on every temp utility). Obviously there is an issue, but what it is I just don't know.

Though I will checkout 3dmark and get back to you.

Thanks

 

[bobwya]

If you had posted those screenies last year we could have cleared this up then!! This is (as stated already) just a software bug (in Everest). You can't have a 532 Mhz FSB for one die and 375Mhz FSB for the other!! Come on that is common sense surely!!

Got any objections to CPUz??

Bob

 

[zforce67]

If you had posted those screenies last year we could have cleared this up then!! This is (as stated already) just a software bug (in Everest). You can't have a 532 Mhz FSB for one die and 375Mhz FSB for the other!! Come on that is common sense surely!!

Got any objections to CPUz??

Bob

I wasn't here last year, but anyways, its not just an issue with "Everest", as the temperatures of cores 3 and 4 suggest. CPUz and all other utilitys for some reason or another don't show cores 3 and 4, basically it is locked on 1 core, thus the reason we don't have a screenshot of cores 3 and/or 4 using CPUz or an other utility.

Though, all utilities present the temperature and its always the same for cores 3 and 4, a steady 43c-again, no matter the utility.

Thanks

 

[zforce67]

Here is a screenshot with various utilities:



Interesting thing here is, PC wizard shows cores 3 and 4 to be running at a slower FSB.

Odd.