Recent Conversation with MSI and Neo4 overclocking "thing"

Question

I had a recent email transaction with MSI if anyone is interested. I tried my best to represent the group, and I know we all experience the problem a little bit different which is why it has been hard to explain and fix. I tried to give them the best info to duplicate the issue in the lab. It seems clear at this point that MSI still isn't duplicating the issue. They are using 166 Mem Clock Index, and the HT freq of 3x and still not understanding the issue, hopefully I have gave them a guide to duplicate what we have been trying to say. Yes I have explained this to MSI before but I really feel I am getting somewhere with my last 2 emails. Once we get them to duplicate the results then I think we will all have our solution shortly. Before anyone says anything negative, at least they aren't saying, oh that's an overclocking problem we don't deal with that. Also remember how long it has taken to get many people to see the issue, this isn't easy to understand so just stay cool, don't say negative sh-t if you reply and we'll see how this pans out. It was actully kind of hard since I am sitting in a hotel room 500 miles away from the motherboard but I was able to use modbin6 to view the BIOS and since I have done this about 7 zillion times I remember the settings LMAO. They have their Venice to 2750, not bad... 
Hello MSI Engineer,
The information you requested. The data you sent shows a Venice running at 166 Memclock index value at HT at 3x, the goal is to run at 200 Memclock Index and HT at 4x. In order to duplicate this issue in the lab here is my recommendation. It MUST be tried with a CBBID Winchester RevD CPU and a Venice Rev E CPU and the Gskill memory you have in dual channel. The issue happens everytime with a CBBID Winchester and less often with Venice.
Starting with the Venice 3500+, Set the following in BIOS
[Memclock Index] 200
[CAS# latency (Tcl)] 3
[Min RAS active] 11
[RAS to CAS delay] 6
[Row Precharge] 6
[User config mode] manual
[bottom of 32 bit] (default)
[1T/2T] 1T
[Read Preamble] 6ns (default)
[Async Latency] 6ns (default)
[S/W mem hole] Disable (default)
[H/W mem hole] Disable (default)
[MTRR mapping] Continuous (default)
[High Performance] manual
[Agreessive timing] Disable
[Dynamic Overclocking] Disable
[Adjust CPU FSB Frequency] 200
[HT Frequncy] 4
[HT Width] 16 up 16 down
[CPU Spread] Disable
[SATA Spread] Disable
[PCI Spread] Disable
[PCIe Clock] 101
[SSE/SSE2] Enable
[CnQ] Disable
[CPU Ratio] 11x
[Adjust CPU VID] 1.45
[Extra CPU Voltage] your choice for stability
[Mem Volt] your choice for stability
[NF4 Volt] your choice for stability
Now boot into Windows. Using Clockgen, increase HTT to 250. This should be stable based on your screen shot and your RAM modules. Now having a CPU 2750Mhz, RAM at 250Mhz, and a HT of 4x250(1000). Test for stablity with Prime95 or other Stress Testing tool. Now we can say motherboard is stable at 2750Mhz CPU, 250Mhz RAM, and HT 4x250(1000). Now return to the BIOS at set the following
[Adjust CPU FSB Frequency] 250
This reults in a no POST condition for me and others. Now install the CBBID, except you will have to set a lower CPU max speed based on CBBID overclocking potential. In the BIOS, Return [Adjust CPU FSB frequency] to 200, Perhaps [CPU ratio] of 10x or 9x but a [Memory Clock Index] 200Mhz. Boot to Windows and use Clockgen to raise HTT to 250 again. Once you find a [CPU Ratio] that will have a STABLE overclock with a 250HTT. Return to the BIOS and set [Adjust CPU FSB Frequency] to 250. Try and POST. Try this experiment with the 3.1, 3.3, and 3.4 BIOS's. For the CBBID and the 3.1 BIOS you should be able to change the [HT Freq] to 3x, 2x or 1x and get a POST with [FSB Freq] 250, [Memclock Index] 200, and a known stable [CPU Ratio] you found in Windows. The 3.3, and 3.4 BIOS have made this work around not possible. Why does [HT Freq] 4x work in windows but not at POST time?
As a side note. The 1.3 BIOS for the 7125 works with the Venice, and the 1.4 and 1.5 do NOT. So somewhere between the 1.3 and 1.4 MSI created the issue on the Venice. Perhaps that is where you can find the answer and solution to the problem.
Here are some articles that describe the issue on a ECS and Chaintech board that were remidied by a BIOS update. So the problem is not uncommon and should be able to be fixed.
http://www.xbitlabs.com/articles/mainboards/display/chaintech-nforce4_6.html
http://www.xbitlabs.com/articles/mainboards/display/ecs-kn1-extreme_9.html
Here is some other users data of NF4 SLI and Ultra MSI Neo4's if you want to see some other BIOS settings.
https://forum-en.msi.com/index.php?topic=81364.0
Here is a screen shot of my Neo4 Plat SLI (MS-7100) 3.1 BIOS and CBBID Winchester(The Venice will NOT POST at all above [Adj CPU Freq] ~219-229 under any situation but is stable in Windows at HTT 250 with Clockgen). The system is stable and POST's when [HT Freq] is 3x, it will NOT at [HT freq] 4x which is stable in windows but won't POST.
http://home.comcast.net/~aicjofs/cbbidMSI2501.jpg
Here is another screenshot of the Asus A8N Plat SLI with all peripherals the same. Only changed the motherboard. This is POSTing at [HT Freq] 4x. This is just as a reference that the components do indeed POST with [Adjust FSB freq] 250, [Memclock Index] 200, and [HT Freq] 4x. This board will POST both my 3200+ Venice and Winchester CBBID at the settings described above.
http://home.comcast.net/~aicjofs/cbbid250asus1.jpg
The following are my BIOS settings that will create a non-POST condition, but the board is stable at these settings when ADJUSTED FROM WINDOWS.
[ADVANCED BIOS]
[Gate A20] Enable
[APIC] Enable
[MPS Table] 1.4
[Boot OS2] No
[ADVANCED CHIPSET]
[System BIOS Cacheable] Disabled
for newer BIOS
[SLI Brodcast apeture] Whatever default is.
[INTEGRATE PERIPHERALS]
[Onboard IDE 0 and 1] Enabled all channels in Auto
[IDE DMA] Enabled
[SATA 1,2,3,4] Enabled
[IDE PREFETCH] Enabled
[NVDA RAID] Enabled
[all IDE raid channels] Disabled
[all SATA raid channels] Enabled
[On Chip USB] 1.1-2.0
[USB KB/Storage] Disabled
[NVDA LAN] Enabled
[NVDA ROM] Disabled
[Onboard VIA 1394] Enabled
[Marvell Gigabit] Disabled
[Onboard Creative] Enabled
[SiI3132] Disabled
[FDC] Enable
[Serial Ports] Disabled
[Parallel Port] Disabled
[POWER MANAGMENT] all defaults
[PNP/PCI]
[Init display] PCIe
[Resources] Auto
[PCI Express Max Payload] 4096
[CELL MENU]
[Timing mode] manual
[Memclock index value(MHz)] 200
[CAS# latency (Tcl)] 3
[Min RAS active] 8
[RAS to CAS delay] 4
[Row Precharge] 4
[Row cycle time] Auto
[Row refresh] Auto
[Row to Row delay] Auto
[Write Recovery] Auto
[Write to Read] Auto changing this to known working value will lock system sometimes
[Read to Write] Auto changing this to known working value will lock system sometimes
[Refresh] Auto
[User config mode] manual
[bottom of 32 bit] whatever it defaults to
[1T/2T] 1T
[Read Preamble] 6ns
[Async Latency] 6ns Attempts to change this to 8ns will lock system
[S/W mem hole] Disable
[H/W mem hole] Disable
[MTRR mapping] Continuous
[High Performance] manual
[Agreessive timing] Disable
[Dynamic Overclocking] Disable
[Adjust CPU FSB Frequency] 250
[HT Frequncy] 4 (For 1000Mhz)
[HT Width] 16 up 16 down
[CPU Spread] Disable
[SATA Spread] Disable
[PCI Spread] Disable
[PCIe Clock] 101
[SSE/SSE2] Enable
[CnQ] Disable
[CPU Ratio] 10x
[Adjust CPU VID] 1.45
[Extra CPU Voltage] +9.9%
[Mem Volt] 2.7v
[NF4 Volt] 1.65v
BIOS Tested 3.0, 3.1, 3.2, 3.3, 3.4. I have not tested 3.51 but forum users are still not getting positive results
-------------- Original message --------------
Dear Aicjofs
This is MSI technical support engineer from Taiwan.
We receive some message from MSI LA office & Anandtech.com.
We are sorry for inconvinence experience when you try to Overclock MS-7100.
We hope to contact with you directly and help you to solve it.
Depends on related message from others, we know that you have several problem for MS-7100 overclocking.
According to your info at MSI Forum, Here is your system configuration:
Neo4 Plat SLI(MS-7100) 3.1 BIOS
3200+ Venice (8x313) @ 2500Mhz
Patriot PC4000 (200:133) 2.5-3-3-7
HT x3
Aopen 6800GT @ Ultra speed
AthenaPower 500W 3.3v - 26a 5v - 47a 12v - 28a
Could you give us more info about what changed item inside your BIOS?
[Memclock index value(MHz)]
[CAS# latency (Tcl)]
[Adjust CPU FSB Frequency]
[HT Frequncy]
[Adjust CPU VID]
[Extra CPU Voltage]¡Kor others
We try to Overclock MS-7100 with BIOS v3.2 and 3500+ (Venice core).
CPU ADA3500DAA4BW ABBWE made in Malaysia
Memory G-skill PC4400 512MB*2 (Dual Channel)
We only change some item inside BIOS as: (Based on Optimized defaults)
[Memclock index value (MHz)] à166
[CAS# latency (Tcl)] à 3
[Adjust CPU FSB Frequency] à250
[HT Frequncy] à3x
[Adjust CPU VID] à 1.450V
[Extra CPU Voltage] àOver VID 6.6%
We will try to find similar CPU as yours to test it.
If possible, please tell us the CPU OPN number like we listed. (ADA3500DAA4BW)
We also help to know any other info that you can provide.
If possible, please help to tell us more setting and system info.
Thanks!
Waiting for your responding.
Best Regards!
Technical Support Engineer
Technical Support Department
E-mail: msi.com.tw
[Post edited by Glenn]

Answer 1

Quote from: Indigian on 15-June-05, 19:41:43
Good news and good work guy's
On a side note to chakkasol....
are you using coolbits 1 or 2?
I heard that coolbits2 is the one for our later cards
Good question - coolbits 2.  I happened to take a chance from this website and downloaded this file...ran it up against my antivirus and antispyware software and it looked clean.  I simply ran the program and viola, coolbits functionality automatically added to the winxp registry.  I cleaned out the msi DOT drivers first before I loaded this program.  Even though I have tweaked on occasion the winxp registry to speed things up and make winxp shut down quicker, this was too complicated for me to do on my own.
The highest overclock with the MSIs DOT is 11% at 555/1100.  My MSI card is actually very stable at 580/1160 for a 16% overclock.  I have gotten it to 600/1200 but I usually run the coolbits stability program a few times and it has never failed at 580/1160.  I am sure if I had a better cooling system than the stock fan it would probably go to 600/1200 or higher.  At a 3DMark05 of 4133 (and much higher if I could get it stable at 600/1200, I would be bumping up against the stock speeds of the 6800s vid cards.   
EDIT: Of course, if you download this coolbits 2 program you do so at your own risk...I take no responsibility for the download.
The coolbits download:
http://downloads.guru3d.com/download.php?det=815
The MSI 6600gt card that I purchased:
http://www.zipzoomfly.com/jsp/ProductDetail.jsp?ProductCode=321916
btw, I have ordered more stuff from ZZF than newegg...they seem cheaper (but dont always have the stock of newegg) and the stuff gets here 1 day sooner usually (west coast to east coast USA).