Lost 2k
- Thread starter fatty
- Start date
maverik-sg1
New member
2k oh my god!!
Just on a side note - try running 14x250 or 13x265 - the extra bandwidth should enhance your scores at the expense of best overclock.
Well here's the thing, did you do a complete re-install or just swap boards? Or did you re-install over your exisiting installation? Try re-installing the nforce4 drivers and perhaps the video drivers?
Also, disable system restore and change the virtual memory drive to D (defualt at C).
Plus - if you cards are not fully above 35degrees you may be experiencing some cold-bug issues - try warming them up first (switch pump of for a minute and run with no rad fans connected).
Just a few ideas.
Mav
Just on a side note - try running 14x250 or 13x265 - the extra bandwidth should enhance your scores at the expense of best overclock.
Well here's the thing, did you do a complete re-install or just swap boards? Or did you re-install over your exisiting installation? Try re-installing the nforce4 drivers and perhaps the video drivers?
Also, disable system restore and change the virtual memory drive to D (defualt at C).
Plus - if you cards are not fully above 35degrees you may be experiencing some cold-bug issues - try warming them up first (switch pump of for a minute and run with no rad fans connected).
Just a few ideas.
Mav
maverik-sg1
New member
fatty said:
Try re-running the test - I will also take some piccies of my BIOS set-up (BIOS 623-x) which you can use.
Cheers
Mav
Phil Stanbridge
New member
Did you download the NF4 chipset drivers or did you install the ones of the CD? Download new ones if you havent already done so.
maverik-sg1
New member
Pateience my young Padawan, patience.
It's more likely that as the Wendy board is a later release than the DFI that the disk you had was the latest one and the BIOS was too.
The DFI is a mature board - there is every reason to think that you need the BIOS updating and the drivers supplied on the disk are the oldest ones available.
Update to BIOS 623-1 (for TCCD) and use the latest Nvidia NF4 chipset drivers.
All should be less depressing and lots more fun
It's more likely that as the Wendy board is a later release than the DFI that the disk you had was the latest one and the BIOS was too.
The DFI is a mature board - there is every reason to think that you need the BIOS updating and the drivers supplied on the disk are the oldest ones available.
Update to BIOS 623-1 (for TCCD) and use the latest Nvidia NF4 chipset drivers.
All should be less depressing and lots more fun
maverik-sg1
New member
If u did not live so far away I'd come and set it up for ya - maybe scorchio can help ya if he's down that way?
On a side note - Betty has finished her tour of duty and you can be proud.
Will be back at Rear10 tomorrow mate.
Cheers
Mav
On a side note - Betty has finished her tour of duty and you can be proud.
Will be back at Rear10 tomorrow mate.
Cheers
Mav
just a quick question - I thought that with the new HT memory controller on the Athlons it meant that higher FSB or HTT base speeds were becoming less and less important? e.g. the gain from 200Mhz to 250Mhz isn't all that much, neither is 250 to 300Mhz. Overall clock speed is still king.
N
nick25
Guest
name='chris_ah1' said:
Possibly getting confused with memory dividers?
Actually no - the person nipped it earlier - I forgot to say that going to a higher FSB and having to sacrifice latency, especially 1T. Although a sacrifice of cas2 to cas2.5 is no worry for the majority of overclockers (of course not those looking to be top 10 
)
Although, now you mention memory dividers I guess why not run at 300HTT if you can? Then run your memory at a speed where you don't have to sacrifice latency. But still, how many percent is in it seeing as you guys have the experience? Hopefully more than the error in benchmarks, and hopefully more than just an academic difference.
And yes, before you ask the nubbishness over the practicalities of HTT is becuase I'm still in socket A world with a 2500+ mobile.
)Although, now you mention memory dividers I guess why not run at 300HTT if you can? Then run your memory at a speed where you don't have to sacrifice latency. But still, how many percent is in it seeing as you guys have the experience? Hopefully more than the error in benchmarks, and hopefully more than just an academic difference.
And yes, before you ask the nubbishness over the practicalities of HTT is becuase I'm still in socket A world with a 2500+ mobile.
N
nick25
Guest
chris_ah1 said:Actually no - the person nipped it earlier - I forgot to say that going to a higher FSB and having to sacrifice latency, especially 1T. Although a sacrifice of cas2 to cas2.5 is no worry for the majority of overclockers (of course not those looking to be top 10
Although, now you mention memory dividers I guess why not run at 300HTT if you can. Then run your memory at a speed where you don't have to sacrifice latency.
And yes, before you ask the nubbishness over the practicalities of HTT is becuase I'm still in socket A world with a 2500+ mobile.
Again because generally bandwidth outweighs latency, it does depend on specific numbers though i.e how much extra bandwidth is worth sacrificing latency.
Not to be too persistant - but does anyone have any benchmark runs done to actually show that the difference in overall performance from that extra bandwidth in let's say 3dmark05 for simplicity's sake is more than 3-5%? That's what i find more interesting. Or better yet - which apps would really show the benefits of 10x300 over 12x250 over 15x200 (given you used dividers to keep memory at the fastest clocks at which they can maintain their lowest latencies).
Forgive me, but this socket939 overclocking seems so much more interesting.
Forgive me, but this socket939 overclocking seems so much more interesting.
There is a small thread http://xtremesystems.org/forums/showthread.php?t=55318&highlight=cas2 on there about it
