Asus P5W DH with GSKILL ram setup instructions and discussion thread

[Mick22]

Dont know, truth be told im not that savvy with overclocking
But 4-4-4-12 @ 889 looks very good.

[lapdog]

There may be a BIG difference from 3:5 and 1:1.

[Polynikes]

Well... it seems OCing the CPU did cause some problems. After upping the FSB to 333mhz x 10, but leaving the RAM at DDR2-888 both SuperPi and SP2004 Orthos give me errors. I tried relaxing the timings to 5-5-5-15 but that didn't help. However, Orthos only gives me errors on tests that stress memory; the CPU-only tests have no problems. Unfortunately Orthos says the errors are logged to a stress.txt file which apparently doesn't get written, because I can't find it.

In both cases I get rounding errors, with SuperPi it happens within 10 seconds, sometimes 10 sometimes just one. In Orthos it happens almost immediately. What could be the problem? The RAM is running at the same speeds, latency and voltage as before with no errors, but now they happen really quickly. Any ideas on how to make it more stable? The CPU OC isn't very aggressive, and I'm running it at 1.3v in the BIOS, (both CPU-Z and Core Temp display different voltages) which I think should be plenty of juice, unless I'm mistaken about that.

[Mick22]

Did you try leaving the ram @ 5-5-5-15 and setting it to ddr-667? Unless im missing something.

[Polynikes]

No, it's perfectly stable at SPD settings. (DDR2-800) I just don't understand why at DDR2-889 with the CPU at stock speeds it's stable but when I OC the CPU to 333mhz FSB with the RAM still at DDR2-889 I start getting errors on tests that stress RAM.

[Mick22]

It does sound odd. But it is possible that the overclock will be stable at a lower ram setting(667), why? I dunno

[Polynikes]

It is rather odd. I put the CPU back to stock speed and everything is running smoothly (no errors) at DDR2-889 with 4-4-4-12 timings. Stock speeds with the x6800 are fast enough for me, so I'm not gonna worry about it. When RAM prices eventually go down I'll get some faster sticks. (And if RD600 turns out as good as the hype says I'll probably grab one of those.) This NQ stuff was cheap so even getting this modest overclock with faster timings is good enough.

[Mick22]

That is very true and i thank you for checking out its overclockability
Ill most likely just overclock my 6600 to ~3Ghz with 667 settings and that will be that unless i can find a higher stable setting which i doubt but for that price it isnt really that much of a problem

[Polynikes]

The NQ sticks' SPD is DDR2-800, so you shouldn't have any problem running them at that speed. (Or around that speed, the FSB speed changes the options available for RAM speed.)

[Thasp]

I have the F2-6400PHU2-2GBHZ ---DDR2 1000

Using those settings, I can't get them to run memtest at all. The RAM is set to DDR890, CPU is at 3.52 GHz, FSB 445 on the e6400. the CPU is prime stable there but the memory can't do a think there. I tried 5-5-5-15 with it too, and it still can't do a thing at that speed.

Is this because this retarded board lacks subitimings?

I tried booting into windows and running memset. I turned the timings up, TRRD to 9, and the other timings up high. First slowly, then put them all as loose as possible, and hit apply, then save, but it still won't pass a second of memtest in windows this way. Is there a way for me to see if it even applied the higher memory settings?

[Thasp]

I tried 440 MHz on RAM, 440-445 FSB, 2.3 volts on the RAM, and loosened almost every setting in memset.. trd 10, etc. Everything was loosened, and it froze after a millisecond of windows memtest.

[Polynikes]

That's strange... perhaps there's a problem with the board or there's something wrong that's common to G.Skill RAM? I got errors as soon as I started any sort of memory test with the CPU overclocked as well, but games ran fine.

[Thasp]

I'm guessing it's a motherboard issue.

[Polynikes]

I guess it doesn't like benchmarks in some circumstances. I never had any noticeable stability issues when I had both the CPU and RAM OC'd, despite the instantaneous errors I got in SuperPi and Orthos. :\

[Thasp]

Errors in orthos or superpi means it's not stable.

I can have 333 FSB and the settings he gave, and be stable.. but 333 FSB is useless. That's a 400 MHz overclock, not a 1.4-1.5 GHz overclock like I have now.

[Polynikes]

Well I guess in my case it's just ty RAM. Strange, though, because I don't get errors running the memory at the same speed but with no overclock.

[Thasp]

I have a theory that it is the motherboard. When the DFI lanparty comes out, I'll confirm this.

[Polynikes]

Yeah, I guess it would be the mobo, since the RAM tests fine at the same speeds with no CPU OC. Strange indeed.

[monza1412]

what about this ones: F2-6400CL4D-2GBPK ??
I want to know if it also suffers of the spd bug with this mobo in particular.

[Scarybeast]

Hi G.SKILL,

I wonder if you can tell me something about an issue I have.

Just recently I bought a 2GB pair of F2-6400CL4D-2GBPK. I am running these sticks on my ASUS P5W DH Deluxe, PCB version 1.03, together with a C2D E6600. My BIOS version is 1602.

The first thing I wonder about is, that actually the Vspec for this RAM on your homepage is 2.0V~2.1V on the german subpages (http://gskill.de/product_sp2630.html), but 1.9V~2.0V on the english subpages (http://www.gskill.com/en/f2-6400cl4d-pk.html). As the stickers on my RAM say 1.9~2.0V, I tend to believe that this is the speced Voltage for my pair of RAM.

However, as I am part of some BOINC projects, I need a 100% stable system, which I always check with Prime95. If my system is running 12+ hours stable with Prime95 I think my system is stable. The experience I made with other stability check apps (like Orthos, which should be running the prime algorithm, or SuperPi) is actually, that my system passes these tests but still can fail on the Prime95 stability test.

Everything on my system is running at stock speeds, dividers and voltages. The first settings I applied to my 2GBPK was 4-4-4-12 (4 tWR) @ 1.9V, as you suggest in the beginning of this thread. At these settings, my system only was running primestable for about 1.5 hours. Then I got the first rounding error in Prime95. To be sure that the RAM is actually not bad I was running Memtest86 for several hours, and it revealed no errors. So I went back to SPD timings (5-5-5-15), leaving the voltage up at 1.9V, and this time my system passed the Prime95 test. Next, I raised the voltage to 1.95V, again reducing the memory timings to 4-4-4-12, and also there my machine was not primestable for more than 1.5 hours. First when I raised the Voltage of the RAM to 2.0V (still within spec) I got it running primestable for more than seven hours.

To be absolutely sure that my system is stable, I tend to apply 0.05V more to the RAM than it needs to pass the Prime95 test. This would be 2.05V here (which my BIOS doesn't offer), so I would end up with 2.1V, which is out of spec according to the lables on my RAM sticks.

Currently I am running a primestable test on 5-5-5-15@1.8V, just to see, if the sticks can make it at defaults.

Do you (or somebody else) have an idea or similar experience? I mean I expected the RAM to be primestable at 4-4-4-12@1.9V... Maybe I got a pair of "not 100% good" RAM sticks?

Thanks already for your answer :-)


--
Intel E6600 @ stock speed @ stock Vcore
ASUS P5WDH 1602 BIOS
G.Skill 2x1GB F2-6400CL4D-2GBPK (currently at 5-5-5-15@1.9V)
ASUS EAX1950PRO

[monza1412]

Hi G.SKILL,

I wonder if you can tell me something about an issue I have.

Just recently I bought a 2GB pair of F2-6400CL4D-2GBPK. I am running these sticks on my ASUS P5W DH Deluxe, PCB version 1.03, together with a C2D E6600. My BIOS version is 1602.

The first thing I wonder about is, that actually the Vspec for this RAM on your homepage is 2.0V~2.1V on the german subpages (http://gskill.de/product_sp2630.html), but 1.9V~2.0V on the english subpages (http://www.gskill.com/en/f2-6400cl4d-pk.html). As the stickers on my RAM say 1.9~2.0V, I tend to believe that this is the speced Voltage for my pair of RAM.

However, as I am part of some BOINC projects, I need a 100% stable system, which I always check with Prime95. If my system is running 12+ hours stable with Prime95 I think my system is stable. The experience I made with other stability check apps (like Orthos, which should be running the prime algorithm, or SuperPi) is actually, that my system passes these tests but still can fail on the Prime95 stability test.

Everything on my system is running at stock speeds, dividers and voltages. The first settings I applied to my 2GBPK was 4-4-4-12 (4 tWR) @ 1.9V, as you suggest in the beginning of this thread. At these settings, my system only was running primestable for about 1.5 hours. Then I got the first rounding error in Prime95. To be sure that the RAM is actually not bad I was running Memtest86 for several hours, and it revealed no errors. So I went back to SPD timings (5-5-5-15), leaving the voltage up at 1.9V, and this time my system passed the Prime95 test. Next, I raised the voltage to 1.95V, again reducing the memory timings to 4-4-4-12, and also there my machine was not primestable for more than 1.5 hours. First when I raised the Voltage of the RAM to 2.0V (still within spec) I got it running primestable for more than seven hours.

To be absolutely sure that my system is stable, I tend to apply 0.05V more to the RAM than it needs to pass the Prime95 test. This would be 2.05V here (which my BIOS doesn't offer), so I would end up with 2.1V, which is out of spec according to the lables on my RAM sticks.

Currently I am running a primestable test on 5-5-5-15@1.8V, just to see, if the sticks can make it at defaults.

Do you (or somebody else) have an idea or similar experience? I mean I expected the RAM to be primestable at 4-4-4-12@1.9V... Maybe I got a pair of "not 100% good" RAM sticks?

Thanks already for your answer :-)


--
Intel E6600 @ stock speed @ stock Vcore
ASUS P5WDH 1602 BIOS
G.Skill 2x1GB F2-6400CL4D-2GBPK (currently at 5-5-5-15@1.9V)
ASUS EAX1950PRO

I´ve your same specs, and I´ve no orhtos problem at 4.4.4.12. But I have inestability problems if I give 2.1 or more vdimm, I mean windows crashes badly. By the way this sticks sucks in the overclocking departament. Best timming I´ve got was 880 at 5.5.5.15, and with the vdimm problem I´m kind of limited here.
By the way, I can confirm that the mobo spd bug beyond 380 fsb is not present with this memory, I was able to boot with manually adjusted timmings.

[GSKILL TECH]

i am sorry about this problem
the voltage is really best run at 1.9v 4-4-4-12-4-42-3-11-7-13.
for memory test we always suggest to run 2 windows memtest with 1000 and 2nd one on all unused memory. if you can run over 1000%+ stable that is great. also 2 super pi 32m is great tool to check. sp2004 you can only said "all system stable test". most the time SP2004 is not just ram depended.

those IC doesnt like high v-dimm. please don't add over 2.2v
if not stable let me know

GST

[Scarybeast]

i am sorry about this problem
the voltage is really best run at 1.9v 4-4-4-12-4-42-3-11-7-13.
for memory test we always suggest to run 2 windows memtest with 1000 and 2nd one on all unused memory. if you can run over 1000%+ stable that is great. also 2 super pi 32m is great tool to check. sp2004 you can only said "all system stable test". most the time SP2004 is not just ram depended.

those IC doesnt like high v-dimm. please don't add over 2.2v
if not stable let me know

GST

Okay, I found some more time to investigate a little bit deeper. Here is, what I found:

On my ASUS P5W DH I can unfortunately not control the subtimings (3-11-7-13) through BIOS. I had a look at Memset, which gives me access to the subtimings, but also not to all of them, at least not according to what the subtimings should be set on a P5B Deluxe.

As far is I got it, the timings should be:

1. 3 - TRRD
2. 11 - Write to Read
3. 7 - Read to precharge
4. 13 - Write to precharge

right?

Memset allows me to edit the last three settings. So I adjusted 2. and 4. to the values you recommended. The 'Read to precharge' can NOT be set to 7 on this board, Memset only gives me 4, 5 and 6 as an options there. So I choose 6.
I can not find the TRRD option at all. There is a tRD, which does not really seem to have anything to do with TRRD.

I have made to window shots from Memset, to see how the BIOS adjusts the subtimings when I choose 4-4-4-12-4 or 5-5-5-15-4 in the BIOS. Here you go:



It seems, that the P5W DH BIOS is setting the 'Write to Read' timing to 10 (not to the 11 suggested by you), as well as it sets the 'Read to precharge' to 5 (not to 7 as suggested by you).

I am currently playing around a little bit with Memset settings to get the sticks running rock-stable on my board at 4-4-4-12-4@1,9V. The reason why they don't seems to be too aggressive subtimings of the ASUS BIOS.

G.SKILL, can you maybe give a suggestion on how to set the timings better, as I do not really have access to set the above timings you suggested?

Thx in advance :-)

[Thasp]

Any idea on the P5W DH issue?

My guess, since the settings here for DDR1000 work at 333 FSB, but are bunk for 375, 400, or 450, is that the P5W DH can't run a high FSB and a high RAM clock at the same time, even if the CPU is stable at the higher clock.. no one seems to know though.

[Polynikes]

That seems to be the case with my RAM/CPU overclock.