8KNXP F6 & Corsair XMS Observations

[ruudpel]

the gigabyte website is saying there are two versions of the GA-8KNXP the Rev. 2.0 one, and the..eh..'normal' one i guess does anybody know the difference? i've only been able to find a difference in BIOS update identifications

[wingzero]

Still running old F5 Bios on a 8KNXP rev 1.0 with DPS2 installed and two XMS3200 512MB modules in Dual Channel mode with no issues since last July. I had some issues with some modules when I bought them and I had to get them exchanged for new ones. I got two failing modules in a row, then the shop finally gave me another working module. MemTest86 3.0 tests passed correctly various times and they're still working so far, although I don't know if they'll pass next summer usual high temps.. :sigh!: Regarding currently sold 8KNXP boards they should all be PCB Rev. 2.0 and that's the reason why the GigaByte website reflects the difference. The newer revision probably adds some better components to better keep up with upcoming Prescott compatibility requirements. If the manufacturer tells you that a Bios is meant for a specific PCB revision then you have to trust it and avoid installing the wrong Bios, it would be the same or worse than trying to use a Bios of a completely different motherboard on yours. Latest 8KNXP Bios release should be F8 if I'm not wrong. I'll soon upgrade and retest with newer Bios. However after some weeks spent testing Corsair memory last summer (check my posts on these forums if they're still there for details) until having got another working module I learned that despite any compatibility issues that might still be there, if a Corsair XMS3200 is fully working then it has to give zero errors even at standard JEDEC settings with no overvolt under MemTest86 3.0. All my tests confirmed this, despite the known compatibility issues affecting some Corsair modules more than others with these boards.. the TWINX seem to have more issues than XMS ones.

[WoodDuck]

Has anyone got any results with the F8 or F9 release BIOS updates from Gigabyte? I have just flashed from F5 to F9 and am trialling now but, unfortunately, have discovered one of my paired TWINX-3200LL chips to be bad, so am in the process of replacing it. I would also appreciate it if someone could also point me in the right direction as to how they are checking specific RAM CAS settings. Is it using the Gigabyte overclock utility supplied with the motherboard, or are there specific programs I should be grabbing???

[Synbios]

You can see/change the settings in the BIOS. Hit CTRL+F1 to see the advanced settings.

[Donald24]

Hi folks, I am wondering, if Gigabyte changed something in the newer revision regarding to the mem-ratios. I currently waiting for the board to arrive, and I would like to oc my CPU with the 1.6 divider. Though the mainboard PDF shows me the possibility of a 1.6 ratio (5:4), all folks here are talking about 1.66, what really would be a 5:6 and 5% less oc. So the PDF is wrong, the 1.6 setting is really 1.666 on the 8KNXP? Or do revisions matter? Don

[Synbios]

We (collectively talking about most of the people who have posted in this thread) have already contacted GB regarding the revision question. The only difference between the revisions is the integrated sound card.

[delloite88]

Just got the 8KNXP MB a few weeks ago and am trying my hand at first self built system. Does anyone know if this supports registered memory? Is the Corsair TWINX1024RE-3200LLPT compatible with the 8KNXP? On startup is beeps and won't pass POST. I pulled the above Corsair memory out and got the system to boot with a spare generic 512mb PC2100 I had around without any problems. Thanks for any help.

[Synbios]

No I dont believe it supports Registered or ECC memory.

[wingzero]

8KNXP is based on i875 chipset so yes, it supports ECC memory modules. I haven't checked i875 specs but I doubt it supports registered modules anyway. I think it can support non-registered ECC DDR modules only.

[DeathAdder]

System Specs: PS: Enermax EG465P-VE (FCA) 430W (pretty sure) CPU: Intel P4 3.2Ghz 800fsb HT Memory: TWINX1024-3200LLPro (512x2) Mobo: Gigabyte 8KNXP Rev 2.0 (Non-Ultra) DPS2 installed BIOS FG Video: nVidia Geforce Ti4600 (soon to be a BFG 5950 Ultra) Sound: SB Audigy 2 ZS Plat w/ LiveDrive HDs: 4xWD 30gig 7200Rpm ATA100 (IDE Raid 0+1) NIC: On-board Intel Pro/1000 CT Roms: 2x Lite-On CDRW(48x and 52x Masters on IDE 1 and 2) Floppy: Generic Misc: DigitalDoc5 Fan Controller No Overclocking attempted or required. Underclocking is unacceptable. BIOS Settings: Reset everything to Optimized defaults, configured the IDE RAID - anything I'm not using is disabled (IE: COM Ports, LPT Port, SATA, USB 2.0 etc). Fresh WinXP Pro SP1 install. Latest and greatest drivers/updates for everything listed above. For the most part, the system seems fairly stable, however: Problem: Running MemTest86 v3.0 Mem stick 1 in slot1,2,3,4,5,6 tested ok Mem stick 2 in slot 1,2,3,4,5,6 test 5 failed Mem stick 1 slot 1,2,3 Mem stick 2 in slot 3,4,5 test 5 failed Mem stick 2 slot 1,2,3 Mem stick 1 in slot 3,4,5 tests ok MemTest86 reporting 1024Meg RAM @ 2563MB/s Sandra SiSoft results attached (a little low?) 3dMark03 results, under 2000pts (horrible IMO) I did a forum search for 8knxp and read every thread, saw multiple suggestions for different timings and voltages. There were so many suggestions and options, it's almost impossible to test them all without going crazy and ripping out the rest of my hair. I know it's been mentioned that the 8KNXP is known for issues with the Corsair Memory. My Questions: 1> Do you recommend a different memory config for this system? Maybe a different Corsair set or a different manufacture all together? 2> Since the one memory stick showed continuous failures, should I have the set RMA'd? 2a> If yes to the RMA - Would you suggest going via Corsair or NewEgg (if I can)? 3> What exactly is that "Top Performance" option in the advanced BIOS settings? 4> Also in the advanced BIOS settings is an option for: CIA: Cruse;Full Throttle;Turbo; etc... what is this, exactly? I do not have another system qualified to test the memory in, nor do I have any other memory to test in this system. I selected Corsair b/c I know their name and know their quality. I particularly liked this set of memory b/c of the black heat sinks and LED indicators (looks nice with the window kit and blue LED fans). I chose the 8KNXP simply b/c it has all the features I require, mainly the IDE Raid (0+1) and I liked the idea of the DPS2 6-phase power system. At the time of purchase, the only other option was a Soyo Dragon something or other, and according is multiple test sites, the 8KNXP out performed the Soyo board in all tests. Performance and stability are a must... instability and BSOD are not optional. I do this work for a living and deal with enough of this crap at work, I don't not need to bring these issues home with me. System operation includes intense gaming, heavy graphics work, image editing, and web site design... Feel free to PM me, e-mail me directly or reply to this thread. Thanks for you time.

[Synbios]

I dont want to jynx myself but I havent had a BSOD or any type of error for months since iv'e updated the BIOS to the F9 (im using rev 1.0) I also run the voltage at +.3v which takes the total voltage up to 2.8v. Memory timings are all default and I did no overclocking. 2.8v is a pretty high voltage for default timings and no overclocking, but as soon as I drop it down to 2.7v or less I get memory application errors. This might be far fetched but I think the motherboard says it is giving more voltage to the memory than what it really is. Its the only thing I can think of.

[wingzero]

I spent many days testing Corsair modules on my 8KNXP system when I got the first defective XMS3200. Since the beginning I had a 100% fully working module that was passing all MemTest86 3.0 tests flawlessly and another one that continously failed. At first I had a cheaper 8IK1100 motherboard and then I decided to switch over to 8KNXP for DPS2 mainly, although I got an ANTEC TruePower 550W on this Chieftech case and so it was already very stable on its own. The defective module kept failing on 8KNXP the same way so I asked the shop to exchange it for me. It was a pain to make them understand and see with their own eyes that one of the two modules was defective, they were testing with a lower quality memory test program running under DOS... and they didn't know what MemTest86 3.0 was... it took me a long boring day to bring my PC there at the shop and test the modules with one of their "technician" which really had no knowledge of IT technology.. one of those kids assembling devices with no theory background at all, I wonder if they'll ever get a clue about system settings and such when they have to configure a customer machine.. Oh well.. I'm no genius and still studying, there's always a lot new to study, but I do have quite some theory background and low-level ImageProcessing,VideoCoding, AudioCoding, MPEG and DSP theory knowledge since I passed my university exams on those subjects... After two exchanges I got another fully working module, at least it has been fully working since last summer....and I hope it won't stop working all of a sudden... What I can tell you is that from all my tests I found out that the non-defective, 100% fully working module kept being fully working on every 8KNXP bios settings condition, either over-volted up to Corsair proposed max guaranteed voltage and at standard JEDEC standard voltage, as well as running flawlessly under MemTest86 3.0 with either JEDEC SPD CAS/RAS settings or Corsair proposed ones. The defective modules were always defective under any condition, either relaxed settings or performance ones. Since you got MemTest86 3.0 errors you can be sure that your data got corrupted here and there. If you did video encoding or zipped any file (especially large ones) or decoded some files then it's highly likely that you got corrupted data in every operation there. It happened to me with a cheap Elixir 512MB DDR PC2100 module on a P4B266 motherboard in early 2002 and I've seen with my own eyes the effect of an apparently fully stable system.. I was using MemTest86 3.0 at the time but since I was getting weird system issues, although perfectly stable and it never trashed, I decided to try MemTest86 and I found out that that module was pure garbage and so I had to replace it. Since then I never install or use any OS if MemTest86 doesn't work perfectly and the system can't pass all tests flawlessly with zero errors. I wasn't using MemTest86 at the time because on my previous Pentium II and III PCs I never got any strange issues and burn-in tests under Windows were always rock solid and I never got corrupted data before. But since I was getting corrupted data on that P4B266 system I had to investigate it further and I found out that the RAM was the only culprit there.

[wingzero]

[QUOTE][i]Originally posted by DeathAdder [/i] HDs: 4xWD 30gig 7200Rpm ATA100 (IDE Raid 0+1) [/QUOTE] The integrated EIDE RAID controller might give you added system issues. If I remember correctly there were some drivers causing large data corruption, issues that later drivers should have solved anyway. I never configured it on my 8KNXP anyway, I'm using an Adaptec SATA 1210SA controller which should use a SiliconImage chipset like the one that's onboard of the 8KNXP but Adaptec provides better customized drivers than SiliconImage generic ones, that's the main reason why I bought the added controller.

[DeathAdder]

Posted By: wingzero [QUOTE]The integrated EIDE RAID controller might give you added system issues. If I remember correctly there were some drivers causing large data corruption, issues that later drivers should have solved anyway. [/QUOTE] I've used EIDE Raid in the past on an Abit TH7-Raid mobo, which used a HighPoint-Tech 370a on-board controller. I'd probably rate that system at about 90%. For the most part, the EIDE Raid config didn't cause many problems... Trying to load Win2k/XP via the F6 method was a serious hassle for me. I always had to use a driver 4 or 5 revisions older to get the OS installed, then update the drivers to the latest version once the OS was up and running. The latest release (2.34) caused me nothing but problems. As far as the 8KNXP "GigaRaid" - I've been fairly impressed with it so far. The setup was easier, the OS install went smooth, etc... no problems at all and I feel fairly confident that this is not the issue. To take it one step further, I'm using MemTest86 v3.0 from a bootable floppy, and I don't think the EIDE Raid even comes into play during these tests. Posted By: wingzero [QUOTE]If you did video encoding or zipped any file (especially large ones) or decoded some files then it's highly likely that you got corrupted data in every operation there.[/QUOTE] It does seem that I'm having some data corruption. Every now and then, XP will perform a Scandisk on a random partition, even though everything was shutdown properly during the previous active operation. I thought this was rather odd. Now granted, we did just talk about the possible EIDE Raid issues, but I am using the latest drivers, latest BIOS etc for the machine and really feel this is not an EIDE Raid issue, but possibly being caused by a faulty memory module. Couple Questions: 1> According to Corsair spec sheet for my TWINX1024-3200LLPro memory, the default/SPD timings are: 2-3-2-6 (Location: [url]http://www.corsairmicro.com/corsair/products/specs/twinx1024-3200llpro.pdf[/url]) is this correct? 2> What's the default voltage? Previous Questions: 3> What exactly is that "Top Performance" option in the advanced BIOS settings? 4> Also in the advanced BIOS settings is an option for: CIA: Cruse;Full Throttle;Turbo; etc... what is this, exactly? I think I will run a few more tests on this set of memory, but I will probably try RMA'ing the set with either NewEgg (If they'll do it, been over 30days) or Carsiar (Oh Ram Guy)... but if I'm going to do that, would you suggest a different set over the TWINX-1024LLPro? Perhaps the 3700 or even the C2 sets? Thanks for you time and informative response.

[Synbios]

According to this: [url]http://www.extremetech.com/article2/0,3973,1151805,00.asp[/url] Top Performance increases the clock rate. I haven't been able to use it, it wont even POST if I enable that thing.

[Beaujames]

Imo the best way to keep your sanity with this board is to just underclock and hope that one of the bios updates will actually fix the problem... and at this point rather buying other memory chips I would just replace the damn board, and save yourself the headache. it also has problems with other manufacturers memery too other than corsair I know this isn't what you want to hear but I have lost too much hair on this board tried too hard and too longyou can find a good 875 board that has most of the same features pretty easily, but this had made me lose all faith I had in gigabyteso I would avoid them in the future.

[DeathAdder]

I stand corrected, previously I stated: [QUOTE]Problem: Running MemTest86 v3.0 Mem stick 1 in slot1,2,3,4,5,6 tested ok Mem stick 2 in slot 1,2,3,4,5,6 test 5 failed Mem stick 1 slot 1,2,3 Mem stick 2 in slot 3,4,5 test 5 failed Mem stick 2 slot 1,2,3 Mem stick 1 in slot 3,4,5 tests ok[/QUOTE] Well, during the dinner hour this evening, I let MemTest86 run for about 2hrs in the last stated Dual Channel Setup (Mem stick 2 slot 1,2,3 Mem stick 1 in slot 3,4,5) In Pass 3,5,8 it reported errors in tests 2,5,5. I do believe I will RMA this set with NewEgg and get a replacement, I'm fairly certian there is a defect with one of these modules.

[wingzero]

[QUOTE][i]Originally posted by Beaujames [/i] [B]Imo the best way to keep your sanity with this board is to just underclock and hope that one of the bios updates will actually fix the problem... and at this point rather buying other memory chips I would just replace the damn board, and save yourself the headache. it also has problems with other manufacturers memery too other than corsair I know this isn't what you want to hear but I have lost too much hair on this board tried too hard and too longyou can find a good 875 board that has most of the same features pretty easily, but this had made me lose all faith I had in gigabyteso I would avoid them in the future. [/B][/QUOTE] Despite whatever Corsair can tell about these issues it's a fact that if you get two modules and one tests perfectly all alone while the other shows errors under MemTest86, like it happened to me, then it's the RAM at fault and not the board. If it was the board design the issue here then it shouldn't work with any modules, instead with the XMS3200 modules I got I was able to see that one was perfectly working under any settings and the other was defective and giving errors despite of motherboard bios settings or dimm slot placement. GigaByte BIOS might have some compatibility issues and some wrong settings BUT this doesn't affect RAM integrity, it simply shouldn't and in fact all my tests demonstrated that. In the end it was a defective set of RAM modules. The truth is that while PC2700 modules are fairly common and cheap to manufacture, anything above that rating, which means PC3200 DDR 400MHz and up are more difficult to manufacture and require higher quality components. Probably to cut on costs all memory manufacturers, not just Corsair, had to buy cheaper components and this is what's causing these kinda issues on the market. I think that now there should be far less defective modules on the market than last summer anyway, at least manufacturing for PC3200+ modules should have improved and the average number of defective ones lowered. Be sure that if a module doesn't pass MemTest86 it's highly likely just that at fault and not the motherboard, the chipset, the dimm slot, the dps2 or the psu... Although using cheaper or too weak PSUs might cause serious stability issues to any P4 system that needs a lot of clean power signal to work correctly.

[Caal]

Death Adder: Which timings did you use for your memtest? Normally one starts with SPD timings and then gradually lowers the settings. I was naive in believing Tom's Hardware when they wrote that this mobo would run Corsair Twinx at 2-2-2-3 timings when I purchased this RAM. If you are concerned about performance see: [url]http://www.gamepc.com/labs/view_content.asp?id=ddr400myths&page=1[/url] This whole low latency thing is a marketing ploy

[DeathAdder]

Posted By: Caal [QUOTE]Which timings did you use for your memtest? Normally one starts with SPD timings and then gradually lowers the settings.[/QUOTE] For the last set of testing, I used the Default/SPD settings in the BIOS (non-tweaked)... which according to Corsair is 2-3-2-6. One stick, no matter which slot, would pass all tests, no questions asked. The other stick, however, would fail continuously. I believe a majority of my issues and low benchmarks are caused by this failed stick and I will attempt to RMA them.

[DeathAdder]

As a side note, I found a link to this document on Gigabyte's website regarding the recommended memory to be used in the GA-8KNXP MB - Unfortunately, it only covers Kingston memory : [URL=http://www.ec.kingston.com/ecom/config/models.asp?Manufacturer=gig]http://www.ec.kingston.com/ecom/config/models.asp?Manufacturer=gig[/URL] Does anyone know if a Corsair version of this type of document exists?

[Caal]

[QUOTE]For the last set of testing, I used the Default/SPD settings in the BIOS (non-tweaked)... which according to Corsair is 2-3-2-6.[/QUOTE] You set SPD in the BIOS. It should be something like 2.5-8-3-3. I only say this because Corsair will ask you to do this before RMA. Also you will get errors at settings that are too aggressive for your set up.

[ironkernel]

I seem to have lucked out in finding this forum. I recently assembled a system much like most in this forum (8KNXP rev 2.0, 3GHz/800 FSB, Corsair XMS (2x512) LL ECC, Radeon AIW 9800 Pro, etc) and I haven't even been able to get the config to even POST. I get a repeating fast beep, which according to Award BIOS is a memory error. Based on some experiences of others that have been down this road - does it sound like the BIOS is the culprit? If so, how is it possble to load new BIOS if I can't boot outside of temporarily using other RAM? Any help would be greatly appreciated.

[Nerces]

As nobody helps me on the other thread, I'm asking the same question here ;): I have a Gigabyte 8KNXP (bought in june so certainly rev 1.x) and bought a TwinX1024-3200C2 to replace my old noname memory but I have big problems with it. My Pentium 4 was overclocked but for the tests with the new modules, I choose to clock it "normally". It is a P4 2.6C so the FSB is 200 MHz. I manually set the memory to 6-3-3-2 at 200 MHz and at first boot Windows crashed about two minutes after the start. After that it couldn't boot properly at all. I ran some MemTest and I get about 8 errors / pass. I decided to do the same test (same timing) at 166 MHz and there were less errors (about 1 / pass) but Windows wasn't very stable (explorer.exe crashes) :(: I check the modules with an Asus A7N8X-E Deluxe (XP2200+) and a Gigabyte GA-8IPE1000-L (P4 2.6C) without any problems at the proper settings (6-3-3-2 at 200 MHz), so I am sure that they are not defective. I also made a bunch of tests and the most annoying thing is that with a lot of settings my computer wouldn't even boot... The only I had to do was to remove the battery to clear the CMOS. I might look a little bit impatient but it's really disappointing to buy very expensive memory and motherboard to get stuck with this kind of bugs... Really disappointed by both Gigabyte and Corsair :[pouts: Please, any kind of help would be much appreciated ;):

[RAM GUY]

I would try to set the timings to By SPD and see if that helps. However, This MB has not performed very well and I have not been able to follow up with Giga-Byte to try to get them to look at this. However, it really looks like a bios or revision issue with the MB to me.