RE: [suse-amd64] Be careful with memory compatibility
Daniele, What motherboard are you using? Typically the RAM compatibility is more motherboard dependant rather than CPU. I have the MSI K8T NEO-FIS2R and CORSAIR 32MX8 (256Meg sticks) for RAM and have had wonderful performance and reliability. -Alain. -----Original Message----- From: Daniele Orlandi [mailto:daniele@orlandi.com] Sent: Thursday, March 18, 2004 7:57 AM To: suse-amd64@suse.com Subject: [suse-amd64] Be careful with memory compatibility I suggest everyone with an Athlon64 (not FX) to thoroughly test his memory with memtest86, expecially with test 5 and 8. I did try with 2 different memory brands without luck. I ended up with Kingston KVR400X64C25/512 (Kingston states that it is ok with Athlon64) clocked at 333 since at 400 it still had problems. I don't know if it is the Athlon64 too much stringent with timings or memories not conformant... but I lost much time trying to find a reliable configuration. Bye. -- Daniele Orlandi -- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help@suse.com
Black, Alain wrote:
Daniele,
What motherboard are you using? Typically the RAM compatibility is more motherboard dependant rather than CPU.
I'm using an Asustek K8V. With Athlon64, however, the memory controller is embedded in the CPU so the motherboard is not involved (except for electrical connections). I switched mainboard and CPU without any change. Bye! -- Daniele Orlandi
Despite memory controller being in the CPU, BIOS still affects memory timing etc. and this is where motherboards differ. Please read the article http://www.tomshardware.com/motherboard/20040112/index.html Regards Sampsa Hario
Black, Alain wrote:
Daniele,
What motherboard are you using? Typically the RAM compatibility is more motherboard dependant rather than CPU.
I'm using an Asustek K8V. With Athlon64, however, the memory controller is embedded in the CPU so the motherboard is not involved (except for electrical connections).
I switched mainboard and CPU without any change.
Bye!
-- Daniele Orlandi
-- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help@suse.com
-- Sampsa Hario Insiders Finland Oy Ab IT Consultant Helsinki, Finland
Sampsa Hario wrote:
Despite memory controller being in the CPU, BIOS still affects memory timing etc. and this is where motherboards differ.
Thanks for the link, it was interesting. However, the whole reason for SPD EEPROM is to tell the memory controller the correct timings and one should expect the memory controller using those timings and work correcly. BIOS settings should be used to "overclock" your memory, not to slow it down... There still is something (the memory controller inside the CPU or the memory itself) that is not following the specifications. I don't know which one it is but I'm beginning to fear it could be the CPU... Bye! -- Daniele Orlandi
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Daniele: On Thursday 18 March 2004 09:55, Daniele Orlandi wrote:
Sampsa Hario wrote:
Despite memory controller being in the CPU, BIOS still affects memory timing etc. and this is where motherboards differ.
Thanks for the link, it was interesting. However, the whole reason for SPD EEPROM is to tell the memory controller the correct timings and one should expect the memory controller using those timings and work correcly.
Hee hee - yes that would be the desirable outcome.
BIOS settings should be used to "overclock" your memory, not to slow it down...
There still is something (the memory controller inside the CPU or the memory itself) that is not following the specifications. I don't know which one it is but I'm beginning to fear it could be the CPU...
It is the BIOS responsibility to configure the DRAM controller in the processor. It should do this based on the capabilities of the RAM (read from SPD) balanced against the capabilities of the particular CPU flavor (140, 242, etc). For example, if you have an Opteron 240 and plug in DDR400 RAM the BIOS should not configure for 400Mhz clock because that processor only supports 333Mhz. An example of doing this wrong were early Tyan K8W BIOS which did this incorrectly and programmed an illegal speed into the DRAM controller. I discovered this the hard way. Don't know if clarified things or just muddied the waters more...
Bye!
Good luck, - Darrell - -- sused@mucus.com "Perfect! ....what am I doing?" -- Washu -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQFAWywKeo6c0kw6mZ0RAhPhAKCQyuehmLs1yz95G45KkuStBEagFwCffw/t AYs3Gsa9Ok/NBIXpnWndbf8= =0D3x -----END PGP SIGNATURE-----
On Thu, 18 Mar 2004 19:14:56 +0200 (EET)
"Sampsa Hario"
Despite memory controller being in the CPU, BIOS still affects memory timing etc. and this is where motherboards differ.
Also the motherboards differ in hardware in how the trace lines from the CPU to the DIMMs are laid out (basically how long the DIMM slots are away from the CPU socket) -Andi
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi Alain: I must concur with Daniele. I have a Tyan s2885 w/dual Opterons and went through a more than a month of poo trying to get this system to work reliably with the OCZ DDR400. Note that this memory was specifically designed to work with Athlon 64 FX. Several RMA's later the vendor ended up changing their SPD programming and the PLL's on the DIMMs. Their tech support was great and it works now - but I really didn't need all that down time. On Thursday 18 March 2004 08:29, Black, Alain wrote:
Daniele,
What motherboard are you using? Typically the RAM compatibility is more motherboard dependant rather than CPU.
That is largely true, but less so with AMD64. The DRAM controller is now built into the processor - it's no longer a function of the core chipset. On the other hand, the BIOS programming of the DRAM controller is very important and is tied to the motherboard. Unfortunately for me, the BIOS on this motherboard does not support tweaking of memory timings.
I have the MSI K8T NEO-FIS2R and CORSAIR 32MX8 (256Meg sticks) for RAM and have had wonderful performance and reliability.
Cool! :-)
-Alain.
Regards, - Darrell
-----Original Message----- From: Daniele Orlandi [mailto:daniele@orlandi.com] Sent: Thursday, March 18, 2004 7:57 AM To: suse-amd64@suse.com Subject: [suse-amd64] Be careful with memory compatibility
I suggest everyone with an Athlon64 (not FX) to thoroughly test his memory with memtest86, expecially with test 5 and 8.
I did try with 2 different memory brands without luck. I ended up with Kingston KVR400X64C25/512 (Kingston states that it is ok with Athlon64) clocked at 333 since at 400 it still had problems.
I don't know if it is the Athlon64 too much stringent with timings or memories not conformant... but I lost much time trying to find a reliable configuration.
Bye.
- -- sused@mucus.com "Perfect! ....what am I doing?" -- Washu -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQFAWddGeo6c0kw6mZ0RApHUAKCUbP3AWbFhL9ip2GESVvp+kK4/gwCfZ6nH 82P974e7GNRu181zXfolMmA= =TSSY -----END PGP SIGNATURE-----
Hi,
I use a SK8N with two Kingston KVR333 and a Opteron 140. It is
possible to tweak memory and clock settings a little without sacrificing
stability. Like FXes, Opteron requires Registred/ECC memory so this setup
could not be a good comparasation system. It is very stable on Suse 9.0
64-bit and with other distros and FreeBSD 5.1
However, if it is possible to change memory settings to less
agressive settings, it could avoid some lock ups. Unfortunatelly this is
not always possible so changing memory maker could help. Also, I'm not sure
but earlier Asus board had bios versions that did not know AMD64-FX, only
Opterons and altought this isn't certain for AMD64, it could help check for
BIOS updates.
Best regards.
(Embedded image moved to file:
pic04169.gif)
(Embedded image moved to file: O: +55 11 F: +55 11 EY/Comm: 4776668
pic02154.gif)Rodrigo De 3523-5494 3523-5540
Vincenzo Monteiro
Darrell Shively
Daniele,
What motherboard are you using? Typically the RAM compatibility is more motherboard dependant rather than CPU.
That is largely true, but less so with AMD64. The DRAM controller is now built into the processor - it's no longer a function of the core chipset. On the other hand, the BIOS programming of the DRAM controller is very important and is tied to the motherboard. Unfortunately for me, the BIOS on this motherboard does not support tweaking of memory timings.
I have the MSI K8T NEO-FIS2R and CORSAIR 32MX8 (256Meg sticks) for RAM and have had wonderful performance and reliability.
Cool! :-)
-Alain.
Regards, - Darrell
-----Original Message----- From: Daniele Orlandi [mailto:daniele@orlandi.com] Sent: Thursday, March 18, 2004 7:57 AM To: suse-amd64@suse.com Subject: [suse-amd64] Be careful with memory compatibility
I suggest everyone with an Athlon64 (not FX) to thoroughly test his memory with memtest86, expecially with test 5 and 8.
I did try with 2 different memory brands without luck. I ended up with Kingston KVR400X64C25/512 (Kingston states that it is ok with Athlon64) clocked at 333 since at 400 it still had problems.
I don't know if it is the Athlon64 too much stringent with timings or memories not conformant... but I lost much time trying to find a reliable configuration.
Bye.
- -- sused@mucus.com "Perfect! ....what am I doing?" -- Washu -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQFAWddGeo6c0kw6mZ0RApHUAKCUbP3AWbFhL9ip2GESVvp+kK4/gwCfZ6nH 82P974e7GNRu181zXfolMmA= =TSSY -----END PGP SIGNATURE----- -- Check the List-Unsubscribe header to unsubscribe For additional commands, email: suse-amd64-help@suse.com ---------------------------------------------------------- The information contained in this communication is intended solely for the use of the individual or entity to whom it is addressed and others authorized to receive it. It may contain confidential or legally privileged information. If you are not the intended recipient you are hereby notified that any disclosure, copying, distribution or taking any action in reliance on the contents of this information is strictly prohibited and may be unlawful. If you have received this communication in error, please notify us immediately by responding to this email and then delete it from your system. Ernst & Young is neither liable for the proper and complete transmission of the information contained in this communication nor for any delay in its receipt. Note: If you have received a delivery failure report, it may be due to the change in the Ernst & Young e-mail domains from "eyi.com", "ey.bm", or "ey.bs" to "ey.com". Could you please make the necessary amendment, if required, and resend the message.
participants (6)
-
Andi Kleen
-
Black, Alain
-
Daniele Orlandi
-
Darrell Shively
-
Rodrigo V Monteiro
-
Sampsa Hario