G
Guest
Guest
Archived from groups: alt.comp.periphs.mainboard.gigabyte (More info?)
I have recently upgraded my system. Pertinent components are:
Gigabyte GA-8KNXP ver. 2.0 w/FH BIOS
Intel P4 2.8C CPU w/Thermalright SP-94 heatsink and 90mm Panaflo fan
1GB (2x512) Super Talent PC4000 (DDR500) memory
nVidia GeForce2 GTS graphics card (obviously not upgraded yet)
Adaptec 29160 U160 SCSI adapter (for hard drives)
BIOS Settings:
CPU Host Frequency - 240.5MHz (962.2 FSB; 3.3677 GHz)
Memory Freq. Formula - 2.0 (which means the memory and the CPU are
synchronous ... 1:1)
RAM - SPD settings (3-4-4-8)
Delayed Transaction - Enabled
PSB Parking - Enabled
Command Per Clock - Auto [disabled]
Fast Chip Select - Enabled
Dynamic Paging Mode - Auto [disabled]
Top Performance - Not enabled
RAM Voltage Control - +0.1v [net 2.6v, the specified value for the RAM
modules]
AGP Voltage Control - +0.3v
CPU Voltage Control - 1.5875v
Power Supply:
+3.3 rail yields 3.44
+5.0 rail yields 4.96
+12.0 rail yields 12.22
According to MBM, the power values stay constant (there is some
variation, but not much).
The system is stable (actually rock stable). The SP-94 heatsink is
awesome (45C max with Prime 95 ... starting at a base temp of 32C);
the CPU has definitely not reached it's overclock limit yet.
The issue I have is with RAM. When I put the system together (I put
the new MB into an existing system ... of course changing the chipset
drivers, etc), it was unstable. Windows XP Pro would randomly crash
(spontaneous reboot, not BSOD). After *much* experimentation (and
sending one of the RAM modules back to the mfr. to be tested; it was
fine), I determined that I could get stability by increasing the AGP
voltage ... and the higher I set the AGP voltage, the more I could
overclock. The RAM voltage is at stock (and anything higher causes
memory errors on MemTest86 and MS WINDIAG); nothing spectacular about
the CPU voltage (which so far has had no effect on overclocking, since
I haven't pushed the CPU to its limit).
Granted, 962.2 FSB with 1:1 isn't bad, but I think it can do better
(or at least the CPU can). The memory should be able to reach a
minimum of 1000FSB ... at least that is the spec.
My surprise here is that *AGP* voltage has been the key to
overclocking. Anyone know why? I know that the AGP card is tied to
memory by virtue of AGP Aperture. I haven't played with any of the
overclock settings for the video card (via the nVidia drivers). My
next upgrade item is a video card (MSI FX 5900XT) ... don't know
whether that will affect my ability to overclock. Unfortunately, I
can't turn the AGP voltage up any higher (it is at the max ... +.3),
so unless I can solve the AGP issue, I'm at the end of the line.
Comments?
Thanks
I have recently upgraded my system. Pertinent components are:
Gigabyte GA-8KNXP ver. 2.0 w/FH BIOS
Intel P4 2.8C CPU w/Thermalright SP-94 heatsink and 90mm Panaflo fan
1GB (2x512) Super Talent PC4000 (DDR500) memory
nVidia GeForce2 GTS graphics card (obviously not upgraded yet)
Adaptec 29160 U160 SCSI adapter (for hard drives)
BIOS Settings:
CPU Host Frequency - 240.5MHz (962.2 FSB; 3.3677 GHz)
Memory Freq. Formula - 2.0 (which means the memory and the CPU are
synchronous ... 1:1)
RAM - SPD settings (3-4-4-8)
Delayed Transaction - Enabled
PSB Parking - Enabled
Command Per Clock - Auto [disabled]
Fast Chip Select - Enabled
Dynamic Paging Mode - Auto [disabled]
Top Performance - Not enabled
RAM Voltage Control - +0.1v [net 2.6v, the specified value for the RAM
modules]
AGP Voltage Control - +0.3v
CPU Voltage Control - 1.5875v
Power Supply:
+3.3 rail yields 3.44
+5.0 rail yields 4.96
+12.0 rail yields 12.22
According to MBM, the power values stay constant (there is some
variation, but not much).
The system is stable (actually rock stable). The SP-94 heatsink is
awesome (45C max with Prime 95 ... starting at a base temp of 32C);
the CPU has definitely not reached it's overclock limit yet.
The issue I have is with RAM. When I put the system together (I put
the new MB into an existing system ... of course changing the chipset
drivers, etc), it was unstable. Windows XP Pro would randomly crash
(spontaneous reboot, not BSOD). After *much* experimentation (and
sending one of the RAM modules back to the mfr. to be tested; it was
fine), I determined that I could get stability by increasing the AGP
voltage ... and the higher I set the AGP voltage, the more I could
overclock. The RAM voltage is at stock (and anything higher causes
memory errors on MemTest86 and MS WINDIAG); nothing spectacular about
the CPU voltage (which so far has had no effect on overclocking, since
I haven't pushed the CPU to its limit).
Granted, 962.2 FSB with 1:1 isn't bad, but I think it can do better
(or at least the CPU can). The memory should be able to reach a
minimum of 1000FSB ... at least that is the spec.
My surprise here is that *AGP* voltage has been the key to
overclocking. Anyone know why? I know that the AGP card is tied to
memory by virtue of AGP Aperture. I haven't played with any of the
overclock settings for the video card (via the nVidia drivers). My
next upgrade item is a video card (MSI FX 5900XT) ... don't know
whether that will affect my ability to overclock. Unfortunately, I
can't turn the AGP voltage up any higher (it is at the max ... +.3),
so unless I can solve the AGP issue, I'm at the end of the line.
Comments?
Thanks