[darkinners]

I have a 2600k overclocked to 4.5G @ 1.28v
and when running LinX, Super Pi etc, the benchmark was obviously slower than other similar setups.

Brief information of my rig
Intel 2600k
Corsair H70 cooler (with 2 Gelid Wing 12 PL)
Gigabyte P67A-UD5-B3
4GB x 4 Corsair Vengeance DDR1600 running @ 1600 9-9-9-24-1T
Asus GTX 480
Intel X25-M 120GB G2 Harddisk
4 SATA 7200RPM Harddisk
Corsair HX850 PSU

now, this is my result(this is 4.7G overclocked, because I lost the capture of 4.5g)

and this is others result(he even use lower end 2500k)

my rig 20% slower than his in the Gflops.

my rig run very stable, no BSOD no Halt no anything after Prime95, Linx, IBT and OCCT stress.

about super PI 1M, I tried to run @4.5G earlier, my result was 8.437
while other people overclock to 4.5, most of them got like 8.1xx

and now at AIDA64 benchmark, in Memory tests, my rig lost to a stock speed 2600HT.

The CPU test my rig was faster than the 2600HT, except one test the CPU
AES

my bios overclock config is like this
BLCK 100.3
Multiplier 45
RAM Voltage 1.54
Vcore 1.28
QPI/VTT 1.2
CPU PLL V 1.74
C1E, EIST and other power management are disabled
Spread Spectrum disable
Ram PC1600 9-9-9-24-1T

this is pretty much what I've configure my overclock.

I've heard when you overclock and your Vcore not enough will cause performance drop, so I tried to increase the Vcore to as high as 1.425v
but the performance is exactly the same.

I am very worry if my CPU or RAM or Motherboard is detective or I did anything not good enough for the overclock,

experienced user can you please help me ? Thanks!

 

[TheBlademaster01]

I used to notice this on my Q9550 too vs others. I guess it's just the RAM...

 

[cory1234]

It's fine. Do you have SP1 installed for Windows 7? That will increase your Gflops.

Also try turning off HT and running Linx

.

 

[Alex132]

Maybe try overclocking your NB?

 

[darkinners]

Quote:

Originally Posted by thealex132;12995393
Maybe try overclocking your NB?

don't know Sandy Bridge has a NB

 

[darkinners]

Quote:

Originally Posted by cory1234;12995363
It's fine. Do you have SP1 installed for Windows 7? That will increase your Gflops.

Also try turning off HT and running Linx

.

yes, my windows 7 64bit ultimate is with SP1 and installed all the updates.
I disabled the HT to run the LinX already, also updated the Linpack Binaries to the lastest in order to support AVX

 

[darkinners]

just ran a Supe Pi 32M test
result :8m8.370s

also significantly slower than others have the same CPU and same clock

 

[Ghostleader]

@darkinners

What´s your setting regarding "TurboPower Limit" and Core Current Limit?

Check so that it doesn´t throttle down during your LinX runs.

Quote:

Originally Posted by thealex132;12995393
Maybe try overclocking your NB?

He probably mean that should overclock your "IMC" (memory controller), he obviously don´t know how the Sandy Bridge works, you can´t overclock the IMC separately with Sandy Bridge, it runs the same as the core.

 

[darkinners]

Quote:

Originally Posted by Ghostleader;12995511
@darkinners

What´s your setting regarding "TurboPower Limit" and Core Current Limit?

Check so that it doesn´t throttle down during your LinX runs.

He probably mean that should overclock your "IMC" (memory controller), he obviously don´t know how the Sandy Bridge works, you can´t overclock the IMC separately with Sandy Bridge, it runs the same as the core.

I disabled Turbo boost in Bios, so the powerlimit is not selectable, should enable Turbo Boost? Thanks for your answer

I tried enable turbo boost in order to break the wattage limited, I both Watts and AMPs set to max 300 and run the LinX again.

HT off 97 Gflops
HT on 101 Gflops

not much a difference, people told me that would have BIG difference

 

[sp4wners]

Maybe you should turn on HT? As far as I saw you have it off. If You want to campare with CPUs that have HT on you can forget about it

Edit: Sorry didn't saw HT was off on the other CPU.

Edit2: do you really care about synthetic benchmark? LOL... Then good luck with it

 

[Alex132]

Quote:

Originally Posted by Ghostleader;12995511
@darkinners

What´s your setting regarding "TurboPower Limit" and Core Current Limit?

Check so that it doesn´t throttle down during your LinX runs.

He probably mean that should overclock your "IMC" (memory controller), he obviously don´t know how the Sandy Bridge works, you can´t overclock the IMC separately with Sandy Bridge, it runs the same as the core.

Sorry just only overclocked AMD CPU's , in which NB hugely boosts the GFlops

Imma Intel overclocking newb

 

[darkinners]

Quote:

Originally Posted by sp4wners;12995881
Maybe you should turn on HT? As far as I saw you have it off. If You want to campare with CPUs that have HT on you can forget about it

Edit: Sorry didn't saw HT was off on the other CPU.

Edit2: do you really care about synthetic benchmark? LOL... Then good luck with it

I was told to disable HT to run the LinX(because LinPack doesn't support Hyper Threading)
normal use I enable HT and the result I posted except the LinX test, the rest are all with HT on.

actually I am not a benchmark whore, I will just run them when my system newly installed and confirm all hardwares are working as they supposed to.

I now saw the result difference from the others and I started to worry if my hardwares are defective, if so I gotta bring them to service

 

[Ghostleader]

@darkinners

Okey, why I wonder was that those three LinX runs was not tight so therefore I wonder if it was throttle down.

So if it´s not throttle down, there´s somehting else going on, how does it look after, for example a 25 run of LinX, monitor cpu usage during the session.

Are you sure you have the latest Linpack?

Can be that your LinX doesn´t work properly with that Linpack you have, try this one Click here to download the latest version of LinX with updated Linpack libraries , from Intel 4.5GHz & 5GHz LinX Stable Club over at XS.

Or try IntelBurnTest

HT don´t work any good when all the thread are doing similary work as in LinX, higher GFlops with HT off.

 

[darkinners]

Quote:

Originally Posted by Ghostleader;12996401
@darkinners

Okey, why I wonder was that those three LinX runs was not tight so therefore I wonder if it was throttle down.

So if it´s not throttle down, there´s somehting else going on, how does it look after, for example a 25 run of LinX, monitor cpu usage during the session.

Are you sure you have the latest Linpack?

Can be that your LinX doesn´t work properly with that Linpack you have, try this one Click here to download the latest version of LinX with updated Linpack libraries , from Intel 4.5GHz & 5GHz LinX Stable Club over at XS.

Or try IntelBurnTest

HT don´t work any good when all the thread are doing similary work as in LinX, higher GFlops with HT off.

thanks again man
yes I pretty sure I have the latest LinPack
I downloaded off Intel's website.

I actually tried few times 5GB Ram use and 25 runs of linpack
the result is the same(slower than other chips)
I did those test with HT off and I did monitor the CPU source most of the time is 100% on all cores.
I guess the LinX ran fine :" (

 

[darkinners]

Oh man!! I found out what was the cause!
it was my firewire audio interface! totally the last place I would look into

I turned off my firewire audio interface, Boom!

what the hell? why would a firewire decrease a PC performance that much?

I investigate further, it was the firewire driver, I change the driver from
"Texas Instruments 1394 OHCI Complaint Host Controller" to "1394 OHCI complaint Host Controller(Legacy)
then when I turn on my Firewire audio interface, the perfromance still great!
still has a little drop but more like 3%, which I don't care!

Thanks every guys here helped me out !! Thanks again : )

 

[Sinlok]

Background apps can steal some performance. Try running with out CPUs and temp apps ect. Also you're right about leaving HT off I go from 70 to 110~ by disabling it for the avx Linx test. OC your memory more can help your results a lot if they will support higher speeds.

 

[darkinners]

Quote:

Originally Posted by Sinlok;12997332
Background apps can steal some performance. Try running with out CPUs and temp apps ect. Also you're right about leaving HT off I go from 70 to 110~ by disabling it for the avx Linx test. OC your memory more can help your results a lot if they will support higher speeds.

thanks Bro, I got it all sorted out, it was the Firewire driver stole the performance what a hack : (

 

[Ghostleader]

@darkinners

Nice to hear that you sort it out

Btw, I see that you only need 1,2 vcore for 4,7GHz, pretty good if you ask me, looks like you got a nice chip there

I need 1,35 for 4,6GHz with my i5 2500K

 

[ABerd]

Quote:

Originally Posted by Ghostleader;12997419
@darkinners

Nice to hear that you sort it out

Btw, I see that you only need 1,2 vcore for 4,7GHz, pretty good if you ask me, looks like you got a nice chip there

I need 1,35 for 4,6GHz with my i5 2500K

My 2500k needs volts also

4.6GHz @ 1.36v

 

[darkinners]

Quote:

Originally Posted by Ghostleader;12997419
@darkinners

Nice to hear that you sort it out

Btw, I see that you only need 1,2 vcore for 4,7GHz, pretty good if you ask me, looks like you got a nice chip there

I need 1,35 for 4,6GHz with my i5 2500K

Thanks man! I feel so relief now : ), proved my hardwares are working fine, hate to bring stuffs to service : (

about the voltage, CPU-Z was reported the wrong voltage.
I tested run 4.7 stable @ 1.31V
5Ghz stable @ 1.425V

but I thought 4.5 is good enough for me : ) and the temperature is whole lot cooler (I ran LinX with AVX at 5Ghz 1.425V with Corsair H70, some core goes as hight as 86 degrees and acoustic temperature was 19 degrees.
can't stand it when summer comes .

4.5 is pretty fast already I'd say, thanks again man, you made me think out of the box to solve the problem