OPP could you please run some pifast! :) I'm interested in seeing what you'd score @ 3ghz ;)
Thx.
Printable View
OPP could you please run some pifast! :) I'm interested in seeing what you'd score @ 3ghz ;)
Thx.
236 IN DUAL DDR
VDD@2.v
VMEM@2.9v
2*Winbond 5ns
tearing because of vmem mod on the GC (TI 4200)
http://membres.lycos.fr/cpulloverclock79/236dual.jpg
I'm certain it's a software error. Like I said before there are many different ways of detecting cpu speed programmatically. Seems that MBM is also using a method that doesn't work well with the NF2. (IDEA: perhaps the MBM author might be able to shed some light on the cause of the problem?)Quote:
Originally posted by STEvil
I wonder if we can reproduce these "CPU speed" errors..?
I suggest we get like 5 cpu hogging probrams (CPU burn, prim95, sandra burn-in, seti@home for example) and run them all simultaneously...
If this works correctly, it should make the cpu speed do what 3dmark is "experiencing" and we should be able to detect it by hitting refresh with CPU-Z or WCPUID repeatedly...?
Since MBM5 cant detect the cpu speed correctly, but WCPUID can, maybe its a software error?
This sucks. Unless somebody finds a bios setting or something that gets around the problem it seems that many good NF2 scores won't get published. :(
*EDIT* I've posted here on the MBM forum, let's see if we can get any more info.
cpulloverclock:)
Oustanding bandwidth my friend !! :thumbsup:
Also, Welcome to XtremeSystems !!
Randi:D
That is some monster bandwidth. Can you bench 3d with that? Yeesh.
I'm glad I ordered this board. Got the non-plus after newegg had the + and the 8K3A+ that I returned out of stock. Maybe I can get my xp1600 up to 2100mhz. I need to unlock it and push the memory up higher. All these high scores make me want to push the limit even more. :D
Man... that is great! :toast:
And that's at 2.9v? If so... have another :toast: on me. :D
the NF2 isnt the only board it doesnt detect the cpu speed properly on.
3dmark and MBM5 wont detect my cpu properly on a Shuttle AK11 rev 3.2...
Running an 1800+ at 11x150.89..
MBM detects it as: 1.6
3dmark detects as: 1.6
wcpuid detects as: 1659.78/79.
MBM always has detected cpu speed wrong for me as well as 3dmark. The only reliable program in the respect is WCPUID. Even on this kt333 board I run so I dont think it has anything to do with Nforce 2 its the programs.
Off topic here as well sorry
OPP
So which do you like the best as far as Vapo or Prom.
Which is the best for some fast swapping and least worries about condensation.
Well, back to on topic...:D
OPP, are you still trying to get a bench that will publish with this board...or are you waiting for some sort of response from FM...or what is the current status of things? Inquiring minds want to know!
word.Quote:
Originally posted by BrainStorm
Well, back to on topic...:D
OPP, are you still trying to get a bench that will publish with this board...or are you waiting for some sort of response from FM...or what is the current status of things? Inquiring minds want to know!
Quote:
Originally posted by cpulloverclock
236 IN DUAL DDR
VDD@2.v
VMEM@2.9v
2*Winbond 5ns
tearing because of vmem mod on the GC (TI 4200)
http://membres.lycos.fr/cpulloverclock79/236dual.jpg
ouch
now that would run a killer AMD score :D
I agree. Same with my KX7-333.Quote:
Originally posted by STEvil
the NF2 isnt the only board it doesnt detect the cpu speed properly on.
3dmark and MBM5 wont detect my cpu properly on a Shuttle AK11 rev 3.2...
Running an 1800+ at 11x150.89..
MBM detects it as: 1.6
3dmark detects as: 1.6
wcpuid detects as: 1659.78/79.
But the problem with the NF2 seems to be that it detects it wrongly AND the detected speed seems to fluctuate wildly for some reason.
I was thinking that knowing the detection method being used might give some insight. There is one well known method that compares the cpu's internal cycle counter to an external timer (the PC's 8254 RTC timer) over a given delay period and calculates the speed from cpu cycles/time elapsed. But if the RTC goes wobbly (when the chipset is under stress?) ......
A much better idea is to read the multiplier/bus frequency directly from the cpu/chipset if at all possible, which I'm certain is what WCPUID, CPU-Z, etc are doing. Information on how to do this is almost impossible to find though (I've looked!).
Anyway, if I do find out anything else I'll post it.
cpulloverclock, that bandwidth is incredible! :toast: Can you bench with that? :slobber:
237 IN DUAL DDR
2*WINBOND 5NS (ALWAYS)
VMEM 2.9v
VDD 2.22v
http://membres.lycos.fr/cpulloverclock79/237.jpg
:slobber: :slobber: :slobber: :slobber: :slobber: :slobber:
Isn't 2.22v VDD dangerously high on these boards? Don't blow it up! :D
RE OPP's publishing problem: Alex van Kaam, the MBM author, has pointed me to this FAQ which confirms that MBM uses the counter + timer method to measure cpu speed. So if it's consistently showing the wrong speed (only when overclocked I assume?) it indicates that something is going wrong with the timer when the chipset is stressed (Windows uses the 8254 timer for high resolution timing).
So presumably 3dMark is doing the same thing. That's about as far as I can take it though, I have no idea how you might go about fixing this. Perhaps some more cooling somewhere? Which bit of the chipset contains the timer?
Maybe someone would like to try this out on some mobo tech guy (although given that the Epox one doesn't even know if the PCI bus is locked or not, I don't hold out much hope there :D).
There's a guy whose board glew up at 2 V so be careful.
Interessting thing ...Quote:
Originally posted by MrLavender
:slobber: :slobber: :slobber: :slobber: :slobber: :slobber:
RE OPP's publishing problem: Alex van Kaam, the MBM author, has pointed me to this FAQ which confirms that MBM uses the counter + timer method to measure cpu speed. So if it's consistently showing the wrong speed (only when overclocked I assume?) it indicates that something is going wrong with the timer when the chipset is stressed (Windows uses the 8254 timer for high resolution timing).
So presumably 3dMark is doing the same thing. That's about as far as I can take it though, I have no idea how you might go about fixing this. Perhaps some more cooling somewhere? Which bit of the chipset contains the timer?
What do mean with timer? Do you mean Clock Generator? This must be an ic in combi with a quartz I think (so on the kx7). I have had a look at all ic's on the 8rda, but all seems to have other functions.
The 8254 used to be a seperate IC in the original PC's, but nowadays it's part of the chipset since it's such a simple device. Some details here from Intel's site, or search Google for "8254 timer" or "8254 PIT".Quote:
Originally posted by br@insc@n
Interessting thing ...
What do mean with timer? Do you mean Clock Generator? This must be an ic in combi with a quartz I think (so on the kx7). I have had a look at all ic's on the 8rda, but all seems to have other functions.
I don't know enough about mobo hardware to say where it gets the clock signal from (I'm a programmer, not an engineer :D), I know that the input frequency is 1.19318MHz which gives a max timing accuracy of ~838 nanoseconds. This is why it's used for all high resolution timing by the OS, there is no other device in the PC with that sort of accuracy (sigh, the price of backward-compatability).
Ok, but on the KX7 I find an ic just behind the agp slot, it's a Realtek called RTM560. In SoftFSB you can choose this type of ic, so I thought it's the general clock gen
You gonna smoke something man, be aware.... :rolleyes:Quote:
Originally posted by cpulloverclock
237 IN DUAL DDR
2*WINBOND 5NS (ALWAYS)
VMEM 2.9v
VDD 2.22v
:confused: :smileysexQuote:
Originally posted by TodB
You gonna smoke something man, be aware.... :rolleyes:
Quote:
Originally posted by SKATAN
ouch
now that would run a killer AMD score :D
SKatan how is your 8RDA working ?
You'll be looking :confused: sooner rather than later when that board decides to commit suicide. That is putting a lot of strain on the board as a whole and I sure as hell hope you don't plan to keep it there for much longer.
Maybe a bit of a vdimm mod is in order, with 3.0 - 3.2 you could go a little easier on the vdd. Maybe even get higher fsb ;)Quote:
Originally posted by cpulloverclock
237 IN DUAL DDR
2*WINBOND 5NS (ALWAYS)
VMEM 2.9v
VDD 2.22v