PASSMARK 6.1 CPU Speed error

[rogerk]

I have a Core 2 Duo E6600 running at Bus 485, FSB 1940
Multiplier x 7 = 3.4 Ghz (3395)

Shows CPU at 4366 Mhz (I wish)

CPU Manufacturer:GenuineIntel
Number of CPU: 1 (2 Core(s)/CPU, 1 Logical(s)/Core)
CPU Type: Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
CPU Speed: 4366.4 MHz

[David (PassMark)]

The CPU speed is the speed that is actually measured when PerformanceTest (PT) starts up. It is measured again each time you start PerformanceTest. So if you quit PT and restart it, does the number change?

[rogerk]

Im not shure what you mean?

My CPU Multiplier is limited to 7

this gives a maximum CPU speed of 3.4 Ghz at my 485 Bus speed

I ran passmark 6.1 and the report says my CPU speed is 4.3 Ghz (I Wish)

Your program is not alone several other programs miss read the multiplier and assume it is running at the normal 9 X

Roger

Benchmark Results

Test NameThis ComputerCPU - Integer Math294.9CPU - Floating Point Math800.2CPU - Find Prime Numbers864.1CPU - SSE/3DNow!6924.4CPU - Compression7644.2CPU - Encryption46.5CPU - Image Rotation1534.9CPU - String Sorting4882.7Graphics 2D - Lines106.0Graphics 2D - Rectangles74.2Graphics 2D - Shapes50.5Graphics 2D - Fonts and Text247.9Graphics 2D - GUI158.3Graphics 3D - Simple1008.0Graphics 3D - Medium233.2Graphics 3D - Complex64.5Memory - Allocate Small Block2410.2Memory - Read Cached2593.0Memory - Read Uncached2546.9Memory - Write2279.9Memory - Large RAM619.4Disk - Sequential Read106.5Disk - Sequential Write101.7Disk - Random Seek + RW5.9CD - Read4.2CPU Mark1973.52D Graphics Mark455.9Memory Mark772.5Disk Mark774.3CD Mark514.03D Graphics Mark417.8PassMark Rating987.0
System information - This Computer

CPU Manufacturer:GenuineIntel
Number of CPU: 1 (2 Core(s)/CPU, 1 Logical(s)/Core)
CPU Type: Intel(R) Core(TM)2 CPU 6600 @ 2.40GHz
CPU Speed: 4366.4 MHz
Cache size: 4096KB
O/S: Windows Vista (WIN32)
Total RAM:2045.9 MB.
Available RAM:1283.8 MB.
Video settings: 1600x1200x32
Video driver
DESCRIPTION:
GeForce 7600 GT
MANUFACTURER:
NVIDIA GeForce 7600 GT
BIOS:
Version 5.73.2218A63D
DATE:
08/07/06
Drive Letter: C
Total Disk Space: 931.5 GBytes
Cluster Size: 4 KBytes
File system:NTFS
Comments:
Gigabyte P35 DQ6


PassMark(TM) PerformanceTest 6.1 (http://www.passmark.com)
Results generated on: Monday, November 05, 2007

[David (PassMark)]

We don't look at the multiplier. So it impossible we "misread" it. We measure the speed by direct measurement. If you consistently get this result each time you use PerformanceTest then either the real time high precisons timers on your PC are wrong or the CPU really is clocked at 4.3Ghz.

There is also the more remote possibiity that the CPU has a fault or our software has a bug. But as no one else reported this bug in the last few years, and the results match other software, I don't think it is a bug.

[rogerk]

Intel TAT (Termal analisis tool) gets it right , so does CPU Z!

Intel TAT (Termal analisis tool) gets it right , so does CPU Z!

vista gets it wrong as does my toolbar CPU usage meter

I know my CPU is not clocking 4GHZ +, if only (though it is possible to do)

though it is running an 80% bus overclock and a 40% CPU overclock

Its a Gigabyte P35 DQ6 board, with latest Bios F6! there is a beta 45nm bios but have not installed it yet, not got a 45nm CPU yet

Think it is a bug, though it might be reading cpu speed from Vista, and Vista gets it wrong!

I have just added XP 3rd option boot, will try it on this OS

Roger

[David (PassMark)]

We don't read the speed from Vista. We calculate it at a low level by looking at the number of clock cycles per unit time.

There are a few motherboards that have known faults in their PCI to ISA bridge (the southbridge chipset). The faults can cause the high resolution clock to return wrong values. The high resolution clock is also commonly used to calculate actual CPU speed. So this fault could cause bad CPU speed values to be reported (and mess up plenty of other stuff as well).

But there might also be other issues at play.

[rogerk]

Its just seems to me that you are reading the processor stock multiplier not the actual one, it was designed to run at x 9 and because i have overclocked the Bus it is set to x 7!

then the maths work! 485 x 9 = 4365

that is what i expect vista is doing reading the maximum multiplier the processor is capable of and not the multiplier it is running

also the board changes the multiplier depending on CPU load

so at present it steps x6 and x7, CPU z shows both clock speeds as the CPU switches in real time corectly!

[David (PassMark)]

Its just seems to me that you are reading the processor stock multiplier

It might seem that way. But as pointed out, we aren't doing this. We don't calculate the CPU Mhz speed based on the multipler, nor the bus speed. In fact we don't even know these values. So either time is now wrong or the CPU's count of cycles is now wrong. Which might be a MB bug when using specific settings.

[Ian (PassMark)]

This issue is resolved in the latest version of PerformanceTest, v6.1.1013, available here:
http://www.passmark.com/download/pt_download.htm

In this version we have changed the mechanism to calculate the current CPU frequency as newer Intel CPU's (e.g. Core 2 Duo) that have been overclocked by lowering their bus speed multiplier may reflect a speed based on the maximum bus speed multiplier. Note: earlier versions of the Core 2 Duo etc., did not require this new calculation mechanism.

This does not impact the CPU benchmark scores.

Regards,
Ian

[IntelMan]

hello rogerk, sounds like ya did not disable speed step, or c1e.....

that will lock your multiplier in......it wont fluctuate up and down, like

the way ya have it set up now, when idle it drops to x6....and at

full load it will jump to x9.....lock it in at 7 by disabling the 2 things

in the bios...leaving these enabled also can cause the

system to hang.........IntelMan

[Ian (PassMark)]

The change relates to a change that Intel made in their CPU counters, in particular they changed the definition of one of their clock tick counters (to be based on the maximum multiplier, rather than the current multiplier). This is a CPU design change for later Intel CPU's. Hence, we now use a different counter to get the current clock tick number based on the current multiplier.

Regards,
Ian