Overclock.net › Forums › Intel › Intel CPUs › Who can solve this riddle ?
New Posts  All Forums:Forum Nav:

Who can solve this riddle ?

post #1 of 10
Thread Starter 
heya all,


I'm looking for someone who can solve the following riddle for me because I'm lost


I have a computer that has been running fine for several years now:

Asus P6T-WS motherboard
Intel I7 920 @ 2.66 Ghz (stock on air)
6*2GB Kingston DDR3 memory @1333Mhz
MSI GTX 560 Ti Twinfrozer OC graphics
OCZ 600W power supply

I just replaced the graphics card last week by a brand new MSI 780 GTX Ti Gaming OC edition. It gave me 57 fps as benchmark in uniheaven. Because I decided it was time to test an easy overlock I downloaded a program called Turbo V from Asus to change the BIOS BCLK clock setting in windows without the need for rebooting or going into the BIOS

Now here comes the riddle: If I use this Turbo V program in windows to up the BLCK from 133 to 140 and later 150 I see my fps go down consistently with each step when I benchmark again. Out of curiosity I decided to do the opposite and lower the BCLK in steps of 5 until eventually I reached 115. Under that value my pc hangs the second I click "apply" so that seems to be absolute lowest. But for each 5 it dropped I see my fps go up when I benchmark. At the lowest setting my consistent 57fps has gone up to 67-68fps with the same graphics card and this is the only parameter that I change (using the slider in the Turbo V program) so who can explain why this BCLK underclock causes such a major fps gain ??


Extra info:

I just composed a new machine with an Intel I7 4930k @ 4.5Ghz on an Asus P9X79-E WS motherboard and moved that new 780 GTX ti card into this machine. Again the benchmark on the new machine gives me the standard 57 fps but I can't seem to find the setting that makes it go up to 67 fps. That's why I would like to know the exact parameter that affects the video card so much
Edited by profundido - 3/21/14 at 5:44am
post #2 of 10
Put a x5660 on that mobo it`s better upgrade than lga2011
post #3 of 10
Windows has a bit of a problem. Some of the internal system timers are synced with the BCLK at the moment you boot up. If you use software like SetFSB or Turbo V to adjust the BCLK, this will totally screw up the accuracy of the system timers. Programs that report FPS depend on these timers and assume that they are accurate. That's a problem when they are not.

Here's some background info and how you can go about fixing Windows with one simple command.

http://forum.notebookreview.com/gaming-software-graphics-cards/569893-must-read-setfsb-o-c-game-problems-fixed.html
post #4 of 10
Quote:
Originally Posted by unclewebb View Post

Windows has a bit of a problem. Some of the internal system timers are synced with the BCLK at the moment you boot up. If you use software like SetFSB or Turbo V to adjust the BCLK, this will totally screw up the accuracy of the system timers. Programs that report FPS depend on these timers and assume that they are accurate. That's a problem when they are not.

Here's some background info and how you can go about fixing Windows with one simple command.

http://forum.notebookreview.com/gaming-software-graphics-cards/569893-must-read-setfsb-o-c-game-problems-fixed.html

THIS
post #5 of 10
Thread Starter 
very interesting. So basically you say that by doing this my fps wasn't going up in fact lowering even though the benchmark is fooled into displaying the opposite ?
post #6 of 10
FPS is frames per second. If software can not accurately determine how long a second is then it can not accurately measure how many frames per second your benchmarks are running at.

Someone came up with a great idea to sync some of the Windows High Performance timers to the BCLK. This works fine but they forgot to consider what would happen if the BCLK was changed while in Windows. If the BCLK changes compared to the BCLK you booted up at, your computer will not be able to measure time or FPS accurately anymore. This is a known bug so some of the better benchmarking software automatically corrects for this problem but many popular benchmark programs do not.

To fix this problem you need to fix Windows. You need to sync the Windows High Performance timer to a fixed frequency timer. There is a timer in the CPU that runs at a constant 14.318180 MHz. The speed of this timer does not change when you adjust the BCLK so it can provide Windows with an accurate source of data for high performance timing purposes.

bcdedit /set useplatformclock true

This bug still exists but most users have forgotten about it. The modern K series CPUs are mostly overclocked by adjusting the multiplier so you rarely hear about this issue anymore. Even on a new CPU if you adjust the BCLK by only 1 or 2 MHz, you have just screwed up the Windows high performance timer unless you have used the above command and fixed this problem.

If you ever want to go back to timers based on your boot BCLK speed, you can use the following command to return to the default Windows timer.

bcdedit /deletevalue useplatformclock
post #7 of 10
Thread Starter 
oww thx both for mentioning this and thx for the detailed explanation Uncle. I had no idea of this phenomena. I'll redo all my testing in that case with these parameters
post #8 of 10
Thread Starter 
Quote:
Originally Posted by PontiacGTX View Post

Put a x5660 on that mobo it`s better upgrade than lga2011


Interesting remark. So I checked current prices for a x5660 processor and...OUCH ! over 1300€ for that processor ? No thx. I can get a full mobo-mem-cpu upgrade for half that price smile.gif
post #9 of 10
In usa it`s 150/180usd via ebay
post #10 of 10
Quote:
Originally Posted by profundido View Post

Interesting remark. So I checked current prices for a x5660 processor and...OUCH ! over 1300€ for that processor ? No thx. I can get a full mobo-mem-cpu upgrade for half that price smile.gif

http://www.amazon.com/Intel-BX80614X5660-Xeon-X5660-Processor/dp/B003EM81PC

not sure if this deal will translate back to the country your in (or if amazon ships to you)

but a new x5660 for $314.60
My Rig
(10 items)
 
   
CPUMotherboardGraphicsRAM
I5 4670 H81m P33 MSI GTX 760 Hawk G Skill Ripjaws x series 2x4 
Hard DriveOptical DriveOSMonitor
WD blue 1TB Samsung dvd burner SH-224DB/BEBE Windows 7 Asus VN248H-P 
PowerCase
Rosewill Capstone 550w Corsair carbide 200r 
CPUMotherboardRAMHard Drive
Pentium G3220 MSI H81m-P33 G.SKILL Sniper Series 4GB PNY XLR8 SSD 120GB 
Optical DriveOSPowerCase
Cheapest Drive I Was Able To Find Windows 7 Corsair CX 450w NZXT Source 210 
  hide details  
Reply
My Rig
(10 items)
 
   
CPUMotherboardGraphicsRAM
I5 4670 H81m P33 MSI GTX 760 Hawk G Skill Ripjaws x series 2x4 
Hard DriveOptical DriveOSMonitor
WD blue 1TB Samsung dvd burner SH-224DB/BEBE Windows 7 Asus VN248H-P 
PowerCase
Rosewill Capstone 550w Corsair carbide 200r 
CPUMotherboardRAMHard Drive
Pentium G3220 MSI H81m-P33 G.SKILL Sniper Series 4GB PNY XLR8 SSD 120GB 
Optical DriveOSPowerCase
Cheapest Drive I Was Able To Find Windows 7 Corsair CX 450w NZXT Source 210 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Intel CPUs
Overclock.net › Forums › Intel › Intel CPUs › Who can solve this riddle ?