Help with G.SKILL DDR500 (barely overclocked) and Prime95

[5t3v0]

Yes they will do DDR500 but I run my cpu at 280MHz & my memory wont do that. So with the limited divider options on the Epox I have to use the 166 divider which gives a memory speed of 229MHz. I'm hardly likely to sacrifice my cpu overclock just so I can run the ram 1:1 am I? The loss of performance by running memory at 229MHz is negligible compared to the loss in cpu performance I would suffer if I dropped my oc down to 260 say.

Thanks Zim & abaez for the info. Anyone know how TREF is calculated? In my BIOS it is listed as 1.95, 3.9, 7.8 or 15.6 us. How does that calculate to cycles?

 

[Zim]

Originally posted by: 5t3v0Thanks Zim & abaez for the info. Anyone know how TREF is calculated? In my BIOS it is listed as 1.95, 3.9, 7.8 or 15.6 us. How does that calculate to cycles?

In my link above there is a spreadsheet posted. In that spreadsheet are sample settings, including some with Tref in the format you mention.

I got my memory today and copied the settings right out of that spreadsheet. It ran 100% stable at 260MHz right off the bat. I also noticed that moving one value one notch up or down can prevent it from booting.

 

[nealh]

Originally posted by: Zim

Originally posted by: 5t3v0Thanks Zim & abaez for the info. Anyone know how TREF is calculated? In my BIOS it is listed as 1.95, 3.9, 7.8 or 15.6 us. How does that calculate to cycles?

In my link above there is a spreadsheet posted. In that spreadsheet are sample settings, including some with Tref in the format you mention.

I got my memory today and copied the settings right out of that spreadsheet. It ran 100% stable at 260MHz right off the bat. I also noticed that moving one value one notch up or down can prevent it from booting.

It appears from reports by one user who has bought several new sticks that CPUZ SPD=3-5-5-10(GSkill tech at XS this is too allow initial post for mobo) seem to perform less when when OCing....some are having trouble hitting 260..the stick that SPD 3-4-4-8 OCed better

My experience with this same new stick..0603 date from newegg

623-2 bios ..yellow slots seems fine at 260 and 2.6v(seemed ok with one try at 2.5v)....anything above with tinkering and memtest errors out as well as orange slots

704-2BTA bios..orange slots...tried varied of settings at 2.6v in bios and everything at 270 fails....260 fine in fact I will be testing more but I have used Gskill tech setting but 3-4-3-8-1T

IMHO if I could not 270 or 280 no real benefit above 260 anyway becasue you would need a divider, if you are lucky to get a cpu to run at 2700-2800..so the ram will run between 245-250(and you maybe able to tighten to 3-3-3-8 or better)

It seems the newer sticks are not hitting 270+..this check DFI thread where alot of people are reporting trouble

I asked the GSkill tech about OCs on newer stick but he seemed offended..I was not asking for guarantee but what he was seeing....but got UCCC does 260-280

Personally..too many settings to change and do not want to spend a month and end up at same palce if they are pretty much stuck at 260

 

[Zim]

Originally posted by: nealhIt appears from reports by one user who has bought several new sticks that CPUZ SPD=3-5-5-10(GSkill tech at XS this is too allow initial post for mobo) seem to perform less when when OCing....some are having trouble hitting 260..the stick that SPD 3-4-4-8 OCed better

Needless to say, mine is the 3-5-5-10 variety. As I said, I got to 260 with no sweat, so anything above that is gravy. For $150 for 2GB I can hardly complain. I wonder what the 3-4-4-8 guys paid for theirs.

EDIT: I was curious to see how far I could push mine. I got to 280 (using G.Skill Technician's settings) before I got a couple of errors on 2 runs of memtest86. I'm pretty sure I could clear those with a little tweaking, if I could be bothered. Thing is, I would need to hit 300 in order to run 1:1 and that would be pushing it. As it is they are running solid at 270 and 9:10. I'm very happy with that. People are chasing the Corsairs, but they are nowhere near as good as these.

BTW, poor "G.Skill Technician" seems to have gone into hiding with the complaints people are making. Thing is, just because they loosened the SPD doesn't mean they will OC any less... and the fact that some people are not getting big OC's means nothing. Every stick of memory is different, and every guy trying to overclock is different.

 

[nealh]

Originally posted by: Zim

Originally posted by: nealhIt appears from reports by one user who has bought several new sticks that CPUZ SPD=3-5-5-10(GSkill tech at XS this is too allow initial post for mobo) seem to perform less when when OCing....some are having trouble hitting 260..the stick that SPD 3-4-4-8 OCed better

Needless to say, mine is the 3-5-5-10 variety. As I said, I got to 260 with no sweat, so anything above that is gravy. For $150 for 2GB I can hardly complain. I wonder what the 3-4-4-8 guys paid for theirs.

EDIT: I was curious to see how far I could push mine. I got to 280 (using G.Skill Technician's settings) before I got a couple of errors on 2 runs of memtest86. I'm pretty sure I could clear those with a little tweaking, if I could be bothered. Thing is, I would need to hit 300 in order to run 1:1 and that would be pushing it. As it is they are running solid at 270 and 9:10. I'm very happy with that. People are chasing the Corsairs, but they are nowhere near as good as these.

BTW, poor "G.Skill Technician" seems to have gone into hiding with the complaints people are making. Thing is, just because they loosened the SPD doesn't mean they will OC any less... and the fact that some people are not getting big OC's means nothing. Every stick of memory is different, and every guy trying to overclock is different.

You are lucky if you are hitting 280..first I have seen post 280 with the 3-5-5-10 sticks...

I see to have a cold boot issue at 260 3-4-3-8....I am pretty dissapointed my cruical valueram at 225 3-3-3-10 were as fast as these at 260....

 

[Zim]

Originally posted by: nealhYou are lucky if you are hitting 280..first I have seen post 280 with the 3-5-5-10 sticks...

I'm using a MSI motherboard. Maybe that's a factor.

 

[nealh]

maybe so..the dfi can be memory finicky and pain in the arse to setup

 

[OvErHeAtInG]

I just got this ram. Bebop - how did your OC turn out in the end once you got prime stable?

Originally posted by: Zim
Have a read thru this thread.

Thanks for that link Zim. Informative. I've just posted my experiences in this thread.

 

[5t3v0]

I tried the timings from the DFI-Street spreadsheet & lowered some some more but still couldn't get even 260MHz to be stable.

I also tried to run my overclock on 8x315 instead of 9x280 (both = 2.52GHz). It runs ok with the ram on the 133 divider (210MHz) but on the 166 divider (252MHz) its unstable no matter what I do with the timings. Board must be OK at that HTT so I suspect the ram cant do that tiny overclock on a divider.

I'm curently trying CL2.5 on my original OC of 9x280 with the 166 divider (229MHz). After a BSOD I dropped TWTR from 1 to 2 and have been priming for an hour now trouble free.

Timings are as follows:

Hypertransport - 280MHz
HT Multiplier - x3
CPU Multiplier - x9
CPU Frequency - 2520MHz
CPU Voltage - 1.55v
DRAM Divider - 166 (5:6)
DRAM Frequency - 229MHz
DRAM Voltage - 2.6v
Command Per Clock (CPC) - 1T
CAS Latency Control (Tcl) - 2.5
RAS# to CAS# Delay (Trcd) - 3T
Min RAS# Active Time (Tras) - 6T
Row Precharge Time (Trp) - 3T
Row Cycle Time (Trc) - 7T
Row Refresh Cycle Time (Trfc) - 13T
Row to Row Delay (Trrd) - 2T
Write Recovery Time (Twr) - 2T
Write to Read Delay (Twtr) - 2T
Read to Write Delay (Trwt) - 3T
Refresh Period (Tref) - 7.8µs
R/W Queue Bypass - 16x
Bypass Max - 7x
32 Byte Granularity - Disable
DQS Skew Control - Disable
DQS Skew Value - 0
Read Preamble Time - 5.0ns
IdleCycle Limit - 16T
Max Async Latency - 7.0ns
Dynamic Counter - Enable

 

[AznKwaiJai]

Is anyone using Asus with these? I got the 3-5-5-10 ones. I can easily go 250 with these, but going higher is a problem. A8N-E with a venice 3000+.

 

[5t3v0]

Last night's test failed after 2 hours. Prime95 crashed. Better than a BSOD I guess. I have now increased vdimm from 2.6v to 2.7v and its been Prime stable for over 3 hours so fingers crossed. That would be 2.5-3-3-6 @ 229MHz. Nice, but not brilliant for ram thats rated 3-4-4-8 at DDR500.

I get the feeling this ram is rated right at its limit so there's not much headroom in it at all.

Edit. Prime95 just failed after 4 hours. At least no crash this time so getting better all the time. I deserve this CL2.5 seeing as the ram wont overclock for sh!t.

 

[1Dark1Sharigan1]

Reviving an old thread but in any case. I have a set of these coming in next week. Anyone know what's the tighest timings I can run at around DDR480-490?

 

[Zim]

Originally posted by: 1Dark1Sharigan1
Reviving an old thread but in any case. I have a set of these coming in next week. Anyone know what's the tighest timings I can run at around DDR480-490?

It's dependent on your motherboard.

 

[1Dark1Sharigan1]

Originally posted by: Zim
Well my mobo is a pretty good OC (Neo4-F). I'm just hoping I can at least get cas 2.5? Is that unreasonable?

 

[Zim]

Originally posted by: 1Dark1Sharigan1

Originally posted by: Zim
Well my mobo is a pretty good OC (Neo4-F). I'm just hoping I can at least get cas 2.5? Is that unreasonable?

Well, considering it is rated at CAS3 at DDR500, 2.5 at DDR490 may be pushing it - you'll just have to wait and see. The difference between CAS3 and CAS2.5 is neglible, and your board will run happily at either, so I wouldn't worry about it.

 

[Fallengod]

I have mine at 250mhz at 3-4-3-7 timings. Thats as tight as they would do without errors. Ive owned both the OCZ 2gb PC-4000 XTC and now I have the g.skill pc4000. Both use Samsung UCCC chips. They both seem to do the same timings. My OCZ xtc maxed out at 3-4-3-7 @ 250mhz as well. I guess that makes sense considering they both use the same chips. Go figure.

Now I never tried increasing voltage, which ive heard does allow slightly lower timings but.... In any case, as most people will tell you, timings mean little anyways. But expect about 3-4-3-7 at DDR500 speeds. And no, at lower speeds you cant tighten the timings with 2gb pc4000 modules unfortunatly. At 200mhz, they still need cas 3, or error.

Originally posted by: 1Dark1Sharigan1
Reviving an old thread but in any case. I have a set of these coming in next week. Anyone know what's the tighest timings I can run at around DDR480-490?

 

[1Dark1Sharigan1]

Ah, I've read at some other places of people at least getting CAS 2.5 but I guess I'll see how it goes with these arrive. I'll be happy with CAS 3 @ DDR480 as long as I can do 1T.

 

[ctbrown]

I have the 3-4-4-8 variety and these sticks don't like dividers. With 2.6V the max is 267mhz w/o divider (memtest stable and prime stable). And the system craps out using a 9/10 divider with a FSB of 289 for a mem speed of 260mhz (Not memtest stable or prime stable). However, using a 166mhz 5/06 divider with 289 FSB is ok at 2.6V.

Recently I tried raising the volts to 2.7 and that lets me use the 9/10 divider with a FSB of 289 to get a mem speed of 260mhz that is memtest stable but not long term prime stable.

 

[OvErHeAtInG]

Originally posted by: 1Dark1Sharigan1
Ah, I've read at some other places of people at least getting CAS 2.5 but I guess I'll see how it goes with these arrive. I'll be happy with CAS 3 @ DDR480 as long as I can do 1T.

Mine only does up to 230 with 2.5. After that it's 3.