Jump to content

Wildstaroct

Members
  • Content Count

    38
  • Joined

  • Last visited

About Wildstaroct

  • Rank
    New Member
  1. Good thread, was looking for an excuse to try the NV drivers. Based on a couple of tests (boot time, HD Tach), here's my results (switching from MSFT drivers to NV): * Boot time decreased 7.5% to 80 seconds (from 86) * CPU utilization increased 57% to 3.66% (from 2.33) * All other HD Tach stats showed no difference I tested both boot time and HD Tach 3 times and averaged them. So not a big sample by any stretch. Despite the CPU utilization increase (which I find suspect with such a small sample), I'll probably stick with the NV drivers unless a problem starts cropping up. My reasoning: * System feels a bit snappier (totally subjective, I know) * I like faster boots (even if it is only 6 seconds) * I can turn command queuing off on my Raptor 150. I've seen some results suggesting that turning it off in a non-server environment (me) results in increased performance and less potential problems with the NV drivers. The MSFT drivers don't give an option to turn NCQ off. That's it, cheers.
  2. Just DL the latest from DFI's site. It's dated Apr 2006 I believe. http://us.dfi.com.tw/Support/Download/down...jsp?CATEGORY_TY
  3. I'm getting some strange readings w/ the new version of Speedfan, v4.29. In this release, there is a new sensor called AMD K8. The Speedfan change log reads "added full support for AMD K8 (Sempron, Opteron, Athlon 64) internal temperature sensor." This new sensor is reporting load CPU temps 12C higher than the mobo sensor. I've already got the DFI Expert temp offset to +10, so if this new sensor is accurate, it would mean the mobo sensor is off by a whopping 22C. And that I've really been cooking my CPU Any idea which sensor(s) I should believe?
  4. No noticeable change in Obvilion, I tried it in a couple of CPU-constrained areas. No bad side effects tho, tried also in HL2. BTW, I also installed on top of MS hotfix. Edit: Correction: HL2 now seems to fly, much better w/ the fix.
  5. Speedfan is light and effective. http://www.almico.com/speedfan.php
  6. Try setting Vsync prerender limit from 3 to 2. You can do this using either Rivatuner or by enabling coolbits and and clicking the sub-tab under 'performance & quality' settings.
  7. Cool n Quiet and thermal throttling are disabled. Does resuming from standby automatically result in underclocking or is there a way around this? Suppose I could try hibernation...
  8. Anyone know why CPU-Z reports approx 23% decrease in core clock speed and mem freq after resuming from standby in XP? Rebooting brings me back to default speeds, but I use standby so I don't have to reboot! It doesn't appear to be dropping the multi since the numbers don't work out.
  9. If you have an extra fan lying around, mount it to blow on PWN like so. If you've got an extra 3-pin fan header that is controllable by BIOS, even better since you can regulate fan speed. Mine goes about 1000 rpm which isn't even audible. And it reduced PWM temps by 10C+.
  10. Speedfan works fine for me, though I believe w/ the switch you lose some functionality...still has voltage and temp monitoring which is goodness, all I really need. Change your shortcut to this to get it working on the Expert: "C:Program FilesSpeedFanspeedfan.exe" /NOSMBSCAN
  11. Since moving from my Abit IC7 / P4 3.0 Northy to my new DFI Expert / Opty 170, I wasn’t really surprised about how inaccurate s/w temp readings are for CPUs. There have been a couple of threads on this board re: how the Expert underestimates temp, whereas it is common knowledge that the IC7 overestimates temp (reference http://www.bleedinedge.com/reviews/abit_as...s_temps_02.html). While I wasn’t as diligent as the Bleeding Edge folks in testing (e.g., embedding thermal probes in heatsinks), I believe I have an accurate estimate on how much the Expert underestimates CPU temp. And by temp, I mean load temp. Load temps apply to everything below. The answer? Roughly 12C. The ironic thing is that the IC7 overestimates roughly equal to how much the Expert underestimates: 12C. How did I calculate this? By keeping variables constant between systems: • Ambient temp (19C) • Software monitoring tool (Speedfan) • Case temp (28C) • Koolance w/c, same block, same thermal probe location under the block…this is what kicked me off on this effort, the major constant. Both systems registered 29C at load in the Koolance display, but Speedfan registered IC7 at 50C and Expert at 28C. All other factors constant, if we know that IC7 overestimates by an average of 12C, then we can reasonably conclude that Expert underestimates by 12C. BTW, Expert bios used for testing is 2005/12/07. Conclusion: Until more rigorous testing is conducted, or a bios fix is provided (or even possible), Expert users w/ an Opty (170) should add 12C to their load temp to gain a more accurate view of real CPU temp. WildstarOCT
  12. I have both games, no issues. If you're running Doom 3 and Vampire doesn't work, then it's probably a video driver issue. You should get the latest driver (81.98) @ nvidia.com.
  13. Related question - is using Driver Cleaner necessary for chipset drivers? I always use it for video driver updates but wondering if it is advised to use it when updating chipset drivers.
  14. Changing the PCIe bus can have screwy effects on SATA drives. Potential data corruption is not worth the risk of a 1-2% performance increase.
×
×
  • Create New...