PDA

View Full Version : Prime errors with gskill le at only 260!



kickassclone
01-26-2005, 04:48 PM
I am getting rounding errors running my gskill le in prime 95 at 260 htt with a 3xht multi at 2.75 vdimm doing 2.5-3-3-7 1t

This can't be happening WTF I thought this ram was rated at ddr 550 at these timings and it cant' do below stock. This is bs man.

Jupiler
01-26-2005, 04:51 PM
Calm down.

First, try giving it less volts.
2.65V should be enough at 260Mhz.
Run memtest and see if you get any errors with those settings.
If not, then you'll have to look at something else (board, cpu...).

glock19owner
01-26-2005, 04:57 PM
Also...what CPU are you running...prime has issues with some winchesters...

Onepagebook
01-26-2005, 05:27 PM
WTF is this? :confused: (http://www.xtremesystems.org/forums/showthread.php?t=51432)

JNav89GT
01-26-2005, 05:35 PM
I wish people would understand that getting high ram isn't just about the ram. I guess I should just beat my head against the wall though!

JNav89GT
01-26-2005, 05:38 PM
yeah mine sux too!

Karnivore
01-26-2005, 05:49 PM
WTF is this? :confused: (http://www.xtremesystems.org/forums/showthread.php?t=51432)

Curious as to what your trying to say, here OPB??

SIK_L_CELL
01-26-2005, 05:49 PM
Yeah worst ram I ever owned! LOL

http://driverheavenuploads.co.uk/SIK_L_CELL/SI-4.JPG

Karnivore
01-26-2005, 05:53 PM
Well people, I don't see this thread as a brag about "YOUR" Gskill memory thread...


The way I do see it is the thread originator need some help (obviously), so lets lose the sarcasm, and if you cannot control yourselves DON'T POST :stick:

Rather then crap lets help him figure out why....

Onepagebook
01-26-2005, 05:54 PM
Curious as to what your trying to say, here OPB??
what I trying to say is "WTF " is not a good word :D
and I don't see him ask any help after you mentioned it even now.
I mean there are lots of ppl here PM me and ask me help.and I always happy to. it's my pleasure.

Karnivore
01-26-2005, 05:57 PM
what I trying to say is "WTF " is not a good word :D


Ahhh, Gotcha :) and definately agree... that is why the topic has magically changed, and a pm was sent to the thread starter....


Where i was completely confused is by the time you posted, the topic was already changed... this is actually a copy of the original thread as changing topics in Vbulletin is bugged....

Onepagebook
01-26-2005, 05:59 PM
Ahhh, Gotcha :) and definately agree... that is why the topic has magically changed, and a pm was sent to the thread starter....
you are dinfinitely a Good admin, let 's move on and stop all the off topic argue things.:D

Jupiler
01-26-2005, 05:59 PM
Ahhh, Gotcha :) and definately agree... that is why the topic has magically changed, and a pm was sent to the thread starter....

I changed the topic title, on OPB's request. (explained in other topic).


Now, let's get back on topic guys.

kickassclone
01-26-2005, 07:35 PM
Sorry but I did not know about the three letter word problem. I do know now and it won't happen again.

I bought the ram on advice and scores from everyone here and I know the ram is good and that is why I was surprised to get errors running it at 260 at the stock ddr 550 timings. I didn't mean to tick anyone off or badmouth gskill.


I think my proc is topped out at 2.6ghz. I used a divider on the ram and I could boot at 10x 270 but I would get an immediate restart and a bsod. I dropped back down to 260 x10 and went back with a 1:1 ratio on the memory.

My vcore is at 1.65 and my proc is stable but I get rounding errors after 5 minutes of p95 torture test. I upped the vdimm to 2.75 and still the same problem.

What could be wrong? This ram should do the stock timings on 260 like nothing or else I am totally stupid and rounding errors in p95 have nothing to do with memeory.

system specs:

3500+ newcastle
6800ultra not oced
2x512 gskill le
msi neo2 plat 1.4 bios
watercooled by dd tdx
vcore 1.65
vdimm 2.75
htt 260
ht 3x
2.5-3-3-7 1t command rate


Maybe I am just missing something.


FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Wed Jan 26 21:53:54 2005]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

texuspete00
01-26-2005, 07:43 PM
Well, since you are using multi 10, knowing 11 is available I would try to get the same cpu speed with that for testing. I mean it's is more of a cpu test than a memory one. ;)

I would not be surprised... and then there is also the possiblity of your mem controller being at fault. I pretty much bought the GSkills knowing theyve been tested so I would be able to point fingers elsewhere first. It's probably not GSkill failing on it's 275 "garauntee" but quite possibly that AMD never told you it would achieve 2.6GHz nor 260 HTT no?

JNav89GT
01-26-2005, 07:55 PM
down the vdimm to 2.65v at most
down your cpu multiplier to 9x and try to run the HTT up to 275

kickassclone
01-26-2005, 07:57 PM
Well, since you are using multi 10, knowing 11 is available I would try to get the same cpu speed with that for testing. I mean it's is more of a cpu test than a memory one. ;)

I would not be surprised... and then there is also the possiblity of your mem controller being at fault. I pretty much bought the GSkills knowing theyve been tested so I would be able to point fingers elsewhere first. It's probably not GSkill failing on it's 275 "garauntee" but quite possibly that AMD never told you it would achieve 2.6GHz nor 260 HTT no?


Having ram that will do ddr550 means the ram will do ddr550.

Isn't 10x260 the same as 11 x 236

My proc should be able to run at 2.2ghz 275 x8 multi with this ram at the rated timings for the stock speed shouldn't it?

Maybe I just have the whole thing screwed up in my head. I see too many people with the exact same setup as me getting way better results. This is not one or two or three people but hundreds of people. I would really hate to think I am the only one that got all bunk components. I bought the neo2 specifically because it could OC. I didn't buy it to not oc it. If my proc can only run 2.6ghz I can live with that but I did not spend $270 on a gig of ram that won't do rated speeds. I will run it through memtest for an hour or so and if it fails that it's going back.



down the vdimm to 2.65v at most
down your cpu multiplier to 9x and try to run the HTT up to 275


If I do that I won't even be running 2.5ghz but I guess I will know if the ram will do 550.

JNav89GT
01-26-2005, 08:05 PM
Having ram that will do ddr550 means the ram will do ddr550.

Isn't 10x260 the same as 11 x 236

My proc should be able to run at 2.2ghz 275 x8 multi with this ram at the rated timings for the stock speed shouldn't it?

Maybe I just have the whole thing screwed up in my head. I see too many people with the exact same setup as me getting way better results. This is not one or two or three people but hundreds of people. I would really hate to think I am the only one that got all bunk components. I bought the neo2 specifically because it could OC. I didn't buy it to not oc it. If my proc can only run 2.6ghz I can live with that but I did not spend $270 on a gig of ram that won't do rated speeds. I will run it through memtest for an hour or so and if it fails that it's going back.





If I do that I won't even be running 2.5ghz but I guess I will know if the ram will do 550.

the point is to remove the CPU oc as a variable.
You should probably read my G.Skill review before you go blaming the ram. The board could be the problem, the CPU memory controller could be the problem, your oc skills could be the problem etc.... As you say, you see 100's getting good results. Just b/c ram fails memtest doesn't mean it's JUST the rams fault. You need to research and learn prior to blaming parts.
http://www.madshrimps.be/?action=getarticle&articID=244

To RMA the ram might not solve your problem!

texuspete00
01-26-2005, 08:10 PM
Yeah thats my point.... there is a certain possiblity your cpu might fail at 11*236 too. Meaning 2.6Ghz is too much for your newcastle. There are not "tons" doing better than that on air. Mine can do 2.65GHz but I thought that was spectacular and that I got lucky. Lowering the multi to take the cpu out like JNav said is another way... play around.

kickassclone
01-26-2005, 08:42 PM
the point is to remove the CPU oc as a variable.
You should probably read my G.Skill review before you go blaming the ram. The board could be the problem, the CPU memory controller could be the problem, your oc skills could be the problem etc.... As you say, you see 100's getting good results. Just b/c ram fails memtest doesn't mean it's JUST the rams fault. You need to research and learn prior to blaming parts.
http://www.madshrimps.be/?action=getarticle&articID=244

To RMA the ram might not solve your problem!


" G.Skill PC4400-F1-3200DSU2-1GBLE
** Timings used are CAS 2 2:2:5 for 200 MHz operation, and Cas2.5 3:3:7 for 275 MHz Operation. 1T command rate was forced and memory was run at 400 MHz setting in bios which forces memory to scale with HTT in a 1:1 ratio. Memory voltage was set at 2.65v for all testing"

I lowered the multi as you suggested to 9x275 and I booted fine so I decided to try 280 and I booted fine and now I am running p95 at 9x280 with 2.5-3-3-7 1t command rate with no errors so far.

In the above I quoted from your guide, which was very insightful, it says you set the mem divider to 200mhz for a 1:1 ratio. I have mine currently set to "auto". What does "auto" run the memory at?

charlie
01-26-2005, 09:40 PM
See? You always have to eliminate the CPU as a variable first....

C

conrad.maranan
01-26-2005, 10:42 PM
What does "auto" run the memory at?

"Auto" will usually run the memory at whatever its specific SPD states. So, if the memory's SPD is programmed at DDR400, it should default to 200MHz in "Auto." Just set it to 200MHz to be sure you're running 1:1. ;)

kickassclone
01-26-2005, 11:09 PM
Well I did it again!

A nicely corrupted system32 cofig and all 120 gigs of data I had are GONE!

Wowo I just love ocing on a raid array!! Well now that I have a fresh install I might as well go and play with this ram and corrupt it again :banana4: This timne all the important data is not there to lose.

I really get discouraged when this happens. 290 at stock ddr550 timngs and gskill nicely corrupted my windows install so I gather that it ain't doing those timings at that speed.

Well I guess it is back to gathering all the nice progs I had installed and all the movies I had. Probably will take me a few weeks but oh well. I am never using a raid array EVER again as a boot drive. I now have my two wd 160's on seperate drive letters and one of them is only for data. I am not going to corrupt another boot drive filled with all my goodies on a stupid oc mission.

I should have used memtest but that is for another time.

Jupiler
01-27-2005, 12:17 AM
Use the repair function when reinstalling XP.
Happened to me on several occassions too (systems32\config\blablabla).
Rebooted with Xp cd and choosed the repair option (!!! don't choose the first time it's prompts you to repair, but the second time). It will search for previous XP installs and from there, you can repair it.
After that, all should be working fine and no data should be lost.

kickassclone
01-27-2005, 12:39 AM
Use the repair function when reinstalling XP.
Happened to me on several occassions too (systems32\config\blablabla).
Rebooted with Xp cd and choosed the repair option (!!! don't choose the first time it's prompts you to repair, but the second time). It will search for previous XP installs and from there, you can repair it.
After that, all should be working fine and no data should be lost.


I tried that and it never worked I just decided to delete the whole raid array and start over.


The repair option kept sending me to a dos prompt where you could get a list of dos commands and I ran chkdsk on the array and it informed me there was corrupted section that could not be repaired. I tried almost a dozen times trying to save my data but each time I kept getting the same dos prompt and I just got fed up and erased the whole thing.

Now I just have my second drive as a partition where I am going to keep all my movies and prog stored so if I do corrupt again I can just reformat the boot partition and still have all my data safe and sound.

After I finished all my updates and such I went to revert back to the 9x275 at all my old settings and I can't even boot from that configuration anymore. I get kicked back to the bios screen with a prompt telling me to disable all overclocking and only when I do can I reboot again.

Jupiler
01-27-2005, 01:53 AM
The first time you have the option for a repair, just press ENTER, not the "R" key, because it will open a DOS prompt where you, indeed, can open the necessary comands.
After that, it will continue and check for previous installations. If it finds a previous install, choose the correct one (in case of multi boot) and then, you can choose Repair (pressing "R" key). Installation will then continue, and so on...

I never had any problems with a repair, using Raid for almost 2 years now.

glock19owner
01-27-2005, 10:11 AM
Well I did it again!

A nicely corrupted system32 cofig and all 120 gigs of data I had are GONE!

Wowo I just love ocing on a raid array!! Well now that I have a fresh install I might as well go and play with this ram and corrupt it again :banana4: This timne all the important data is not there to lose.

I really get discouraged when this happens. 290 at stock ddr550 timngs and gskill nicely corrupted my windows install so I gather that it ain't doing those timings at that speed.

Well I guess it is back to gathering all the nice progs I had installed and all the movies I had. Probably will take me a few weeks but oh well. I am never using a raid array EVER again as a boot drive. I now have my two wd 160's on seperate drive letters and one of them is only for data. I am not going to corrupt another boot drive filled with all my goodies on a stupid oc mission.

I should have used memtest but that is for another time.

Which SATA's are you using...SATA's 1&2 do not have a working lock so they tend to get corrupted at high FSB...use 3 or 4 if you are not...these SATA's are locked...

Also try using BIOS 1.36b...the extra over volt options do work...and is the only BIOS that has allowed me to benchmark with 5-3-3-2.5/1:1 at 275FB/3X HT

Torin
01-27-2005, 10:32 AM
^^ What he said. Make sure you are using the right SATA ports, 2 are not locked and that will happen everytime.

Like I said in the other thread, you should be testing DDR clock maxes at 8x or 9x multi, to eliminate your CPU limit and discover your mem limit.

And as the Jupiler said, don't use the recovery console, hit enter to install windows, then you will be given the option for automatic repair.

Don't get discouraged! Your memory has a long way to go, but you have to work out the other system issues to get it to max out.

kickassclone
01-27-2005, 11:09 PM
^^ What he said. Make sure you are using the right SATA ports, 2 are not locked and that will happen everytime.

Like I said in the other thread, you should be testing DDR clock maxes at 8x or 9x multi, to eliminate your CPU limit and discover your mem limit.

And as the Jupiler said, don't use the recovery console, hit enter to install windows, then you will be given the option for automatic repair.

Don't get discouraged! Your memory has a long way to go, but you have to work out the other system issues to get it to max out.

I corrupted the system config file. I inserted the xp disk and booted from the cd and went through all the mtions of installing windows from the cd. When it asked to "repair" an install by pressing "r" that is what I did and immediately I was booted to the dos prompt c:> I tried this at least 10 times and everytime it would not repair anything. What could be the problem?

Jupiler
01-28-2005, 01:11 AM
I corrupted the system config file. I inserted the xp disk and booted from the cd and went through all the mtions of installing windows from the cd. When it asked to "repair" an install by pressing "r" that is what I did and immediately I was booted to the dos prompt c:> I tried this at least 10 times and everytime it would not repair anything. What could be the problem?

This is normal if you press "r" the first time. It boots you to the dos prompt.
Don't press "r" the first time it asks you if you wanna repair XP. Just press enter and continue.
It's only the second time you have the option to choose "repair", that you have to press "r". (it will check for previous installations) This will repair you installation, and should normally not boot you back to the dos prompt.

Or does this happen the second time too? Because your post is a bit confusing.

kickassclone
01-29-2005, 12:37 AM
This is normal if you press "r" the first time. It boots you to the dos prompt.
Don't press "r" the first time it asks you if you wanna repair XP. Just press enter and continue.
It's only the second time you have the option to choose "repair", that you have to press "r". (it will check for previous installations) This will repair you installation, and should normally not boot you back to the dos prompt.

Or does this happen the second time too? Because your post is a bit confusing.


If you press enter it installs xp. I do not recall it ever asking if you want to repair a "second" time after you hit enter to install the os.

So what you are saying is that you boot from the xp cd and wehn it asks to install or repair or exit you press enter to install xp. After that it will ask you to repair your install a second time and not go through with the actual installation from pressing enter.

If this is what you mean I only hit hit repair the FIRST time it askes after the initial boot up from the cd. I did NOT press enter like I was just installing the os for the first time. Dam I might have just wasted all my data for nothing if this is the case
:eek:

scottk
01-29-2005, 08:05 AM
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Wed Jan 26 21:53:54 2005]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.

At what FFT size do you get these errors?

shaolin95
01-29-2005, 10:29 AM
The repairs the other poster mentioned is the one that comes when you go ahead like you were trying to install XP again. Dont use the first repair option, select installation and you will be offered another option to repair which doesnt go to DOS and will repair automatically and most of the times it will fix the problems.

saaya
01-29-2005, 10:30 AM
it restarts and then gives you a dos prompt? :confused:
i never heard of that...


dont clock n raid! you should have known better :P

so now whats the status?
the cpu isnt the problem but its the mem?
did you ever get the speed your trying stable with other memory?
ask a friend to lend his memory and see whether it fixed the problem...

skate2snow
01-29-2005, 10:43 AM
I cant affirm that this is 100% true, but it is very possible. MSI K8N Neo2 may have a problem with NewCastle cores.

I have the same problem as you. My BH5 can do 255MHz memtest and on other rigs, but in windows I cant do over 220MHz 2-2-2-7 on this board. And, my WindowsXP Currupts so easily!

I am almost sure that the Neo2 as a compatibility problem with NewCastle. And I'm pretty sure that if MSI knows this problem, a BIOS could repair it.

kickassclone
01-30-2005, 09:16 AM
I cant affirm that this is 100% true, but it is very possible. MSI K8N Neo2 may have a problem with NewCastle cores.

I have the same problem as you. My BH5 can do 255MHz memtest and on other rigs, but in windows I cant do over 220MHz 2-2-2-7 on this board. And, my WindowsXP Currupts so easily!

I am almost sure that the Neo2 as a compatibility problem with NewCastle. And I'm pretty sure that if MSI knows this problem, a BIOS could repair it.


If the board is made for socket 939 and the 3500+ is a socket 939 then why should there be a problem with the 3500+ which is meant to go in a desktop mobo like the neo2. My newcastle cpu works fine I just wanted a better oc outta it than I am getting. I could always just buy an fx-55 but since the fx-57 is gonna be out in the summer I will wait and then have the best amd offers.

kickassclone
01-30-2005, 09:17 AM
At what FFT size do you get these errors?


I get these errors in about 15 seconds after starting p95. These occur almost immediately.

skate2snow
01-30-2005, 09:20 AM
It is not as simple as that, it is a whole coding story.

And this is about the only logical explanation.

bias_hjorth
01-30-2005, 10:39 AM
Do you try the guide from the OS forum? Otherwise its in my sig. for future use :)

mckennaiii
02-01-2005, 06:11 PM
i have tried several times to run prime95 and i get the same errors as in the first post.
my system:
dfi lanparty pro 875b
3.0es
mushkin pc3200 level II [bh5] 4x256
i have all settings set at default.
i have run memtest86 for 60 passes with no errors.
any advice as to why i keep failing prime95 would be appreciated.

FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.
[Wed Jan 26 21:53:54 2005]
FATAL ERROR: Rounding was 0.5, expected less than 0.4
Hardware failure detected, consult stress.txt file.