Dagalidis, about the abit, what is your opinion ? and NF3 vs via ? :p
:D
Printable View
Dagalidis, about the abit, what is your opinion ? and NF3 vs via ? :p
:D
ABIT KV8 Pro is a good mobo but was really hard to make it work perfect...Quote:
Originally Posted by AndreAPL
Was a good performer with low cost and many overclocking options but had hardware problems (VTT Track) which make MEM overclocking limmited to 262 FSB 1:1 using ADATA Vitesta DDR500 with Hynix -D5 chips...
On the other hand DFI until now is the best mobo which i had, and the most easy to overclock....
DFI increased my overclocking potential up to 295 1:1 FSB using the same Vmem compared with ABIT.... :slobber:
Also my CPU reached stable 2500 MHZ using 1.76 Vcore VS 1.82Vcore used on ABIT :D
An other great + on DFI is the fact i can USE HIGH FSB like 357 x 7 = 2500 MHZ @ 277 MEM 1T rock stable...
So YES .... DFI is a great mobo and worth your attention...
here is SS Sandra benches with HTT 4x
Was a reason ever found why some people get better 3D scores with X2 HTT than with X4 HTT?
Bump could anybody answer the last question.
The reason could be that the Hypertransport link becomes unstable at those speeds, yet manages to keep from totally crashing because HT is a packet protocol and can resend packets (a bit like TCP). This causes congestion though, and lowers the 3D scores.
I would try increasing the LDT voltage at speeds > 1000MHz.
HT1.0 doesn't support retry. It doesn't resend errored packets like PCI-E does. Most times when an error occurs on the HT bus the system hangs right away, but not always.
Quote:
Originally Posted by Gogar
i can run 3dmark or any other bench under windows a few times WITH THE SAME SETTINGS and the results are always OFF , 1 or 2 percent plus/minus. Maybe even more.
Therefore i really dont care anymore about a few points up down in any bench under windows....except you do a whole lot of test runs (like 10 or 20) and then average them.
I am always amazed how people give attention to 3dmark/everst etc. scores when they see it's only like 50 or so pts higher w/ a different driver or setting etc.....this means literally NOTHING since there's always the margin up/down where it might differ from run to run anyway.
MAYBE one of the more reliable tests might be rightmark memory analyzer....but then again it also fluctuates SOMEWHAT.