SK8V Memory Help

[Andrew]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

I was wondering if anyone might have an idea of what could be causing
instability problems on the SK8V using 4 Corsair 512MB DIMMS
(CM72SD512RLP-3200)?

The only stable configuration is using 1 dimm in either A1 or A2.
Adjusting the memory speed to DDR333 or DDR200 makes no difference.

Thanks

 

[Ed]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

On Sat, 23 Apr 2005 21:44:48 GMT, Andrew <tireman@shaw.ca> wrote:

>I was wondering if anyone might have an idea of what could be causing
>instability problems on the SK8V using 4 Corsair 512MB DIMMS
>(CM72SD512RLP-3200)?
>
>The only stable configuration is using 1 dimm in either A1 or A2.
>Adjusting the memory speed to DDR333 or DDR200 makes no difference.
>
>Thanks

Have you tried setting Command Rate to 2T?
Ed

 

[Guest]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

I had stability problems with an SK-8V, although I am using CMX512RE-3200XL.
Memtest indicated occasional memory errors. Just a few will make
installation of WinXP difficult, and then it crashes from time to time.
Oddly Suse linux installed and operated without problems. I have had good
success by enabling ECC (Master ECC and DRAM ECC). This does have a
performance cost (~ 10%), but the machine has been solid since.

"Andrew" <tireman@shaw.ca> wrote in message
news:klzae.1120141$6l.918943@pd7tw2no...
>I was wondering if anyone might have an idea of what could be causing
>instability problems on the SK8V using 4 Corsair 512MB DIMMS
>(CM72SD512RLP-3200)?
>
> The only stable configuration is using 1 dimm in either A1 or A2.
> Adjusting the memory speed to DDR333 or DDR200 makes no difference.
>
> Thanks

 

[Paul]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

In article <klzae.1120141$6l.918943@pd7tw2no>, Andrew <tireman@shaw.ca> wrote:

> I was wondering if anyone might have an idea of what could be causing
> instability problems on the SK8V using 4 Corsair 512MB DIMMS
> (CM72SD512RLP-3200)?
>
> The only stable configuration is using 1 dimm in either A1 or A2.
> Adjusting the memory speed to DDR333 or DDR200 makes no difference.
>
> Thanks

There is a thread here:

http://www.houseofhelp.com/forums/showthread.php?p=157133

Down near the bottom, the "RAM Guy" recommends some settings.
Your RAM is rated 3-3-3-8, so substitute those settings.

You might take a look around the houseofhelp site, and post
a question there.

I would start by using the latest beta BIOS. Very few BIOS
have been released for the board, so going with the latest
will give you the benefit of any improvements to the
default settings.

http://www.asus.com.tw/support/download/selectftp.aspx?l1_id=1&l2_id=19&l3_id=3&m_id=1&f_name=1003-007.zip~zaqwedc

AFUDOS flash program
http://www.asus.com.tw/support/download/selectftp.aspx?l1_id=1&l2_id=19&l3_id=3&m_id=1&f_name=AFUDOS211.zip~zaqwedc

One poster in the houseofhelp thread recommends enabling
ECC functions. This is a good idea, but using ECC to fix
flaky memory is not recommended. Instead, you should work on
your memory stability with ECC disabled. If the machine can
be made stable, enabling ECC improves the reliability of the
system, and makes it bulletproof. If you use ECC to hide the
problems, then you lose visibility of how bad your memory
is performing. And the computer can still crash, if an
uncorrectable error occurs.

Increasing RAM voltage might help a bit. At DDR400 rates,
memory should be fed 2.6V (that is the JEDEC spec value).
A little more would not hurt, like 2.7V for example.

The Command Rate setting, which is a popular one with people
using unbuffered memory on other Athlon64 systems, should not
be quite as critical on your system. The reason is, on an Athlon64
system that uses unbuffered memory, two double sided DIMMs on
a memory channel, is 32 loads on the address bus. (That is why
it is called unbuffered, because there are no buffers.) When
registered DIMMs are used, a register chip intervenes on the
address bus. Thus, when two registered DIMMs are placed on a
channel, there are only two loads, which is much easier for
the drivers on the processor based memory controller to handle.

While I haven't read any accounts of testing with or without 2T
command timing on a registered system, I don't see a strong
incentive to kill memory performance by using 2T command rate
timing. (Note - your DIMMs likely have two register chips and
one clock buffer chip - the two register chips work together,
to function as one wide buffer, because one chip doesn't have
enough bus width to get the job done.)

A datasheet for your DIMMs is here - the amazing thing is,
the datasheet spans four pages, but no timing numbers are
shown :-(
http://corsairmemory.com/corsair/products/specs/cm72sd512rlp.pdf

The listing here says the RAM is 3-3-3-8
http://compatible.corsairmemory.com/memorysearch.aspx?modelid=547

The order of these parameters is shown on slide 19 here:
http://corsairmicro.com/corsair/products/tech/memory_basics

HTH,
Paul

 

[Andrew]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Hi Paul,

Thanks for the help. I should mention, as I forgot to in my original
post, I'm running an Opteron 144. I don't know if that makes any
difference or not?

On another note, I think I've had some success. I manually setup the
memory using some of the suggested timings from the Corsair forum link,
setting the voltage to 2.6 and limiting the memory to DDR33. Running
with DDR400 and the above settings would still cause errors. Memtest
would throw one error and the error would cascade repeatedly until the
end of the test. The errors were usually thrown on test 5 or 7.

The current modified settings I have are:

CAS: 3
SDRAM Active to Precharge Delay (tRAS): 6T
SDRAM RAS to CAS Delay (tRCD): 3T
SDRAM Row Precharge (tRP): 3T


For some reason, having the above settings manually set in the BIOS
*seem* to stabilize even at DDR333. Previous attempts at adjusting the
memory speed, low as DDR200, would still cause errors.

Regarding the BIOS, I had tried release 1003.007 from Asus's FTP site
(not listed on the website) and the problem was amplified as the machine
would spontaneously reboot frequently.

Thanks,
Andrew

Paul wrote:
> In article <klzae.1120141$6l.918943@pd7tw2no>, Andrew <tireman@shaw.ca> wrote:
>
>
>>I was wondering if anyone might have an idea of what could be causing
>>instability problems on the SK8V using 4 Corsair 512MB DIMMS
>>(CM72SD512RLP-3200)?
>>
>>The only stable configuration is using 1 dimm in either A1 or A2.
>>Adjusting the memory speed to DDR333 or DDR200 makes no difference.
>>
>>Thanks
>
>
> There is a thread here:
>
> http://www.houseofhelp.com/forums/showthread.php?p=157133
>
> Down near the bottom, the "RAM Guy" recommends some settings.
> Your RAM is rated 3-3-3-8, so substitute those settings.
>
> You might take a look around the houseofhelp site, and post
> a question there.
>
> I would start by using the latest beta BIOS. Very few BIOS
> have been released for the board, so going with the latest
> will give you the benefit of any improvements to the
> default settings.
>
> http://www.asus.com.tw/support/download/selectftp.aspx?l1_id=1&l2_id=19&l3_id=3&m_id=1&f_name=1003-007.zip~zaqwedc
>
> AFUDOS flash program
> http://www.asus.com.tw/support/download/selectftp.aspx?l1_id=1&l2_id=19&l3_id=3&m_id=1&f_name=AFUDOS211.zip~zaqwedc
>
> One poster in the houseofhelp thread recommends enabling
> ECC functions. This is a good idea, but using ECC to fix
> flaky memory is not recommended. Instead, you should work on
> your memory stability with ECC disabled. If the machine can
> be made stable, enabling ECC improves the reliability of the
> system, and makes it bulletproof. If you use ECC to hide the
> problems, then you lose visibility of how bad your memory
> is performing. And the computer can still crash, if an
> uncorrectable error occurs.
>
> Increasing RAM voltage might help a bit. At DDR400 rates,
> memory should be fed 2.6V (that is the JEDEC spec value).
> A little more would not hurt, like 2.7V for example.
>
> The Command Rate setting, which is a popular one with people
> using unbuffered memory on other Athlon64 systems, should not
> be quite as critical on your system. The reason is, on an Athlon64
> system that uses unbuffered memory, two double sided DIMMs on
> a memory channel, is 32 loads on the address bus. (That is why
> it is called unbuffered, because there are no buffers.) When
> registered DIMMs are used, a register chip intervenes on the
> address bus. Thus, when two registered DIMMs are placed on a
> channel, there are only two loads, which is much easier for
> the drivers on the processor based memory controller to handle.
>
> While I haven't read any accounts of testing with or without 2T
> command timing on a registered system, I don't see a strong
> incentive to kill memory performance by using 2T command rate
> timing. (Note - your DIMMs likely have two register chips and
> one clock buffer chip - the two register chips work together,
> to function as one wide buffer, because one chip doesn't have
> enough bus width to get the job done.)
>
> A datasheet for your DIMMs is here - the amazing thing is,
> the datasheet spans four pages, but no timing numbers are
> shown :-(
> http://corsairmemory.com/corsair/products/specs/cm72sd512rlp.pdf
>
> The listing here says the RAM is 3-3-3-8
> http://compatible.corsairmemory.com/memorysearch.aspx?modelid=547
>
> The order of these parameters is shown on slide 19 here:
> http://corsairmicro.com/corsair/products/tech/memory_basics
>
> HTH,
> Paul

 

[Paul]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

In article <nKWae.1127228$Xk.60093@pd7tw3no>, Andrew <tireman@shaw.ca> wrote:

> Hi Paul,
>
> Thanks for the help. I should mention, as I forgot to in my original
> post, I'm running an Opteron 144. I don't know if that makes any
> difference or not?
>
> On another note, I think I've had some success. I manually setup the
> memory using some of the suggested timings from the Corsair forum link,
> setting the voltage to 2.6 and limiting the memory to DDR33. Running
> with DDR400 and the above settings would still cause errors. Memtest
> would throw one error and the error would cascade repeatedly until the
> end of the test. The errors were usually thrown on test 5 or 7.
>
> The current modified settings I have are:
>
> CAS: 3
> SDRAM Active to Precharge Delay (tRAS): 6T
> SDRAM RAS to CAS Delay (tRCD): 3T
> SDRAM Row Precharge (tRP): 3T
>
>
> For some reason, having the above settings manually set in the BIOS
> *seem* to stabilize even at DDR333. Previous attempts at adjusting the
> memory speed, low as DDR200, would still cause errors.
>
> Regarding the BIOS, I had tried release 1003.007 from Asus's FTP site
> (not listed on the website) and the problem was amplified as the machine
> would spontaneously reboot frequently.
>
> Thanks,
> Andrew
>

I have had at least one "untrustworthy" BIOS, that lied about
the DRAM timings. If I use http://www.cpuid.com/cpuz , it was
showing different settings than the ones I was using manually
in the BIOS.

Another tool you could use, to at least observe the settings,
is A64tweaker. I don't know about showing settings on an
Opteron, but since the processors share a lot architecturally,
it is possible it might work.

http://www.akiba-pc.com/download.php?view.40
http://www.xtremesystems.org/forums/showthread.php?s=&threadid=37345

The problem with A64tweaker, and setup of the AMD 64 bit
processors in general, is there is no documentation to correlate
the AMD named parameters, to a memory datasheet. On the private
forums, some people have been playing it by ear and making
rules up as they go. Some of the parameters are determined
by the design of the motherboard (like the length of the
memory bus). But, for starters, i would just fire up that
tool and use it to observe how the hardware is set. There
is no need to tweak anything if you don't want to.

(I don't have an A64 board and haven't tested that app.)

HTH,
Paul

 

[Andrew]

Archived from groups: alt.comp.periphs.mainboard.asus (More info?)

Hi Paul,

Thanks again for all the help. Looks like it was a BIOS setting issue
that was causing the memory instability as everything is stable now.

When I get a chance to try out the Utilities I'll drop and e-mail and
let you know if they worked on the Opteron or not.

Ciao,
Andrew

Paul wrote:
> In article <nKWae.1127228$Xk.60093@pd7tw3no>, Andrew <tireman@shaw.ca> wrote:
>
>
>>Hi Paul,
>>
>>Thanks for the help. I should mention, as I forgot to in my original
>>post, I'm running an Opteron 144. I don't know if that makes any
>>difference or not?
>>
>>On another note, I think I've had some success. I manually setup the
>>memory using some of the suggested timings from the Corsair forum link,
>>setting the voltage to 2.6 and limiting the memory to DDR33. Running
>>with DDR400 and the above settings would still cause errors. Memtest
>>would throw one error and the error would cascade repeatedly until the
>>end of the test. The errors were usually thrown on test 5 or 7.
>>
>>The current modified settings I have are:
>>
>>CAS: 3
>>SDRAM Active to Precharge Delay (tRAS): 6T
>>SDRAM RAS to CAS Delay (tRCD): 3T
>>SDRAM Row Precharge (tRP): 3T
>>
>>
>>For some reason, having the above settings manually set in the BIOS
>>*seem* to stabilize even at DDR333. Previous attempts at adjusting the
>>memory speed, low as DDR200, would still cause errors.
>>
>>Regarding the BIOS, I had tried release 1003.007 from Asus's FTP site
>>(not listed on the website) and the problem was amplified as the machine
>>would spontaneously reboot frequently.
>>
>>Thanks,
>>Andrew
>>
>
>
> I have had at least one "untrustworthy" BIOS, that lied about
> the DRAM timings. If I use http://www.cpuid.com/cpuz , it was
> showing different settings than the ones I was using manually
> in the BIOS.
>
> Another tool you could use, to at least observe the settings,
> is A64tweaker. I don't know about showing settings on an
> Opteron, but since the processors share a lot architecturally,
> it is possible it might work.
>
> http://www.akiba-pc.com/download.php?view.40
> http://www.xtremesystems.org/forums/showthread.php?s=&threadid=37345
>
> The problem with A64tweaker, and setup of the AMD 64 bit
> processors in general, is there is no documentation to correlate
> the AMD named parameters, to a memory datasheet. On the private
> forums, some people have been playing it by ear and making
> rules up as they go. Some of the parameters are determined
> by the design of the motherboard (like the length of the
> memory bus). But, for starters, i would just fire up that
> tool and use it to observe how the hardware is set. There
> is no need to tweak anything if you don't want to.
>
> (I don't have an A64 board and haven't tested that app.)
>
> HTH,
> Paul