790 corruption issues?

[gmcg]

Kinping had NO corruptions, roll back a few pages and see it by yourself.

I am not sure, we know about his testing results only

So the HDDs dont die at all? Anandtech said their (2) 74gb raptors died w/ 790i. As mentioned before, they wouldnt say they were dead if they knew they just had to format them.

I don't understand how the corruption of indexes or files can PHYSICALLY kill HDD.

[eternal_fantasy]

Did you contact Areca or Mtron support?
I suspect that is not casued by the s2e...

If you had to guess which of the following hardware is causing OS corruptions, which one will you put your money on?

The motherboard, which have a 14 page and increasing thread dedicated to the aforementioned problem, or;

The RAID Controller + SSD, that works perfectly on ALL of my motherboards (see sig), but happens to corrupt the OS on the ONE motherboard that are having a 14 page and increasing thread dedicated to the aforementioned problem?

Ever thought that maybe the Striker II Extreme might be the cause?

The corruption is nvidia 570 southbridge related, as usual. doesnt matter if you have a raid controller like Areca or you plug in sata conectors from motherboard, the pci express slots and sata conectors all are controlled by 570.

I thought the first and third PCI-E slot (PCIEX16_1 and PCIEX16_2) are controlled by the 790i SPP?

I installed the Areca on the last PCI-E slot, therefore it may not be caused by the nvidia 570 SB?

I think that many issues could be fixed simply not installing mediashield, and putting the correct settings for memory too, to avoid bsods.

If raising the FSB from 333 to 334 is incorrect settings, and therefore corrupted the OS, then please educate us on what is..
Media shield or not, corruption occurs.

Furthermore, dont you think guys, that current boards still are not prepared for dual channel ddr3 1600 and above? the fsb only reaches 1600 stock, dual channel 1600 is too high for it...

No. Your argument is invalid simply because nVidia advertises their 790i chipset to handle DDR3 speeds of up to and beyond 2000Mhz. Are you sympathizing with nVidia for not able to handle speeds they advertise to support, because they are "not prepared"?

[|SiLA|]

am I the lucky one?i have no corruptions running q6600 g0 @1700fsb 1700mem 7 6 5 18 1T

[saaya]

But that doesnt explain how kingpin and MANY others got their boards 100% stable and some of them with FSB speeds OVER 550Mhz fsb.

Can somebody explain me this cause Im kinda lost here...

i dont think kingpin had NO corruptions whatsoever... thats nonsense...
if you push a system to the limit your bound to get to that point, if you dont see a single BSOD or corrupt a single windows install you havent really pushed a system. :P

But i think there are some boards that definately seem to work 100% fine...
while others dont. I think has to do with the chipst yields, some chipsets can apparently run stable at high speeds while others can not, its as easy as that. many core2 cpus can reach a certain high clockspeed, while only some of them can run it stable as well. thats it, easy as that.

but in this case it seems a bit odd that everybody can reach relatively high clockspeeds but only some can get them stable and some can not. If its just the 570 sb then it should be easy to fix for nvidia... maybe some nvidia sbs suck and some are fine... so thats why some boards work fine and others dont?

But how could the sb cause hdd corruption if people use a pciE raid controller? It should be easy to check then, use a pciE slot controlled by the spp and then one controlled by the sb. if one corrupts while the other doesnt, there you go, its the sb for sure.

but i dont think its the sb... i think the sb ALSO causes corruption and kills some hdds apparently, but i think the main corruption issue is the SPP.
And i actually DO think this can be fixed in BIOS... but im not 100% sure...

especially after nvidias statement... if they knew it could be fixed with a bios then they would have said it. so they either dont know how to fix it or they will fix it in hardware.

or, if they really suck, they will not fix it and just say "too bad? we dont guarantee ocing..."

[saaya]

e_f check the graph, the last 16x slot, the third one, actually IS controlled by the 570 sb!!! so maybe it IS the 570 after all?

could you please try the second pciE slot and see if that makes a difference?

[dinos22]

i don't know if this has been posted here but they are definitely looking into data corruptions resulting from memory and fsb OCs

http://nvidia.custhelp.com/cgi-bin/n...p?p_faqid=2166

[eternal_fantasy]

e_f check the graph, the last 16x slot, the third one, actually IS controlled by the 570 sb!!! so maybe it IS the 570 after all?

could you please try the second pciE slot and see if that makes a difference?

Haha saaya, have a look here;



"The Striker II Extreme has support for both 3-way SLI and Quad-SLI configurations, in addition to the more traditional 2-way SLI setup. The two blue PCIe x16 slots are PCI Express 2.0 compliant and should be used before a third graphics card is installed in the third white PCIe x16 slot (which is PCI Express 1.x compatible only). You can read more about performance tuning for graphic cards in our included 790i BIOS tuning guide." - Anandtech
http://www.anandtech.com/mb/showdoc.aspx?i=3283&p=3

Have a closer read at my post and you will see that I was infact using the second "blue" slot.
Anyhow, I have used both slots 2 and 3, both are affected by the corruption.

(the reference board have the same PCI-E slot layout)

[gmcg]

The problem should be in SPP somewhere. Logically it's performance should be limited by a driver (or a BIOS), therefore if FSB speed is too high to be correctly handled by a CPU or timings are too tight to be handled by SPP/memory, it shouldn't let you boot setting an error flag to BIOS. It doesn't happen and system can't correct or control the errors caused by such situation, sending false commands and data to a HDD and corrupting it.

In BIOS P03 they limited memory bandwidth and latency (and, therefore, it's performance) to make the situation stable enough (it means they have limited the OC abilities of the board, therefore it became to be slower than a regular x38 or x48 non-top over-clocking board).
Now they need to find a better solution, either in BIOS, in SPP driver or in the worst case - in hardware (SPP). They are working on it, as far as we all know, but why should we suffer because of such situation - it is a question.

Well, we still have a solution - we can use seriously limited P03 BIOS with standard timings or find our own limit of stable performance experimenting with FSB and memory timings to have max over-clocking and still have no corruption problems. It's a PITA though. The problem is not so clean also because FSB limitation is different for dual core and quad core CPUs, the first ones support 550 MHz FSB stable enough and the last ones have corruption problems with 450MHz FSB sometimes, therefore the research for limitations requires a lot of patience and ability to restore the corrupted HDDs.

What is worrying me - that Nvidia has found a problem before BIOS P03 release (because they have used bandwidth and latency limitations in it), and they still have no solution.

790i Ultra SLI board is definitely not a good purchase for now, it only can be recommended to people, who like to reach max and don't mind to have some PITA, IMO.

[gmcg]

am I the lucky one?i have no corruptions running q6600 g0 @1700fsb 1700mem 7 6 5 18 1T

Could you post your read-write-copy-latency results and voltage setting , please?

The reason why I ask is if you for example use P03 BIOS, performance of your memory is limited too much to satisfy you with nice timings, but it doesn't mean that you have good results.

Take a look on the following pictures (BIOS P03, no corruption issues):



Settings almost as yours (1800 FSB vs yours 1700, but 7-6-6-15-1T vs yours 7-6-5-18-1T), P03, no corruption issues:



And finally my current setting with BIOS 790P03R2:



Timings don't look so nice, but the memory performance is significantly improved, and still no corruption issues.

[RealTelstar]

If you had to guess which of the following hardware is causing OS corruptions, which one will you put your money on?

The motherboard, which have a 14 page and increasing thread dedicated to the aforementioned problem, or;

The RAID Controller + SSD, that works perfectly on ALL of my motherboards (see sig), but happens to corrupt the OS on the ONE motherboard that are having a 14 page and increasing thread dedicated to the aforementioned problem?

Ever thought that maybe the Striker II Extreme might be the cause?

What is the most obvious may not be the cause sometimes.
I have read somewhere of issues of the mtron SSD drives when used in raid.
my 0.2 cents.

[MateoTTR]

do you guys use system ghost to recover from corruption?
I just don't feel like re-installing windows every time the slightest OC will corrupt it.
Any hint what tricks / software you use for testing corruption, restoring windows, etc.... ??

[sonofander]

I use acronis... There are definately workarounds for this issue. I keep all my files on my non-OS drive and downclock to stock whenever making large file transfers. Whenever I bork my OS (twice now) I do an image restore and everythings fine.

Not that we should have to accept these problems, but really, they're not that difficult to maneuver around...

[saaya]

e_f, well then that settles it for sure, it MUST be the SPP...

The problem should be in SPP somewhere. Logically it's performance should be limited by a driver (or a BIOS), therefore if FSB speed is too high to be correctly handled by a CPU or timings are too tight to be handled by SPP/memory, it shouldn't let you boot setting an error flag to BIOS. It doesn't happen and system can't correct or control the errors caused by such situation, sending false commands and data to a HDD and corrupting it.

what? that doesnt make any sense, why would you implement a mechanism to "check" if a certain speed is stable and only then allowing the system to boot? this doesnt really sound right... are you sure thats how other chipset mfgs do it?

In BIOS P03 they limited memory bandwidth and latency (and, therefore, it's performance) to make the situation stable enough (it means they have limited the OC abilities of the board, therefore it became to be slower than a regular x38 or x48 non-top over-clocking board).
Now they need to find a better solution, either in BIOS, in SPP driver or in the worst case - in hardware (SPP). They are working on it, as far as we all know, but why should we suffer because of such situation - it is a question.

well this sounds like first you release a bios thats tweaked for performance, so then all the reviews look great and some people post great results, but then you release a new bios that actually makes all boards stable but performance sucks... so basically they are selling 2 types of 790 ultra boards, some that can run stable with a tightly tweaked bios and some that cant... come on, wth is that? i honestly pray its not like that...

790i Ultra SLI board is definitely not a good purchase for now, it only can be recommended to people, who like to reach max and don't mind to have some PITA, IMO.

yeah, totally agree... but then again, some boards seem to work perfectly fine even with tweaked bios and high clocks...

i keep hearing about different chipset revisions... b2 and c0?
so the older rev is actually better?
what revisions do you guys have?
c0 seems to be really crappy while b2 steppings seem to work fine?
are there c1 and b1 revisions too?

[truehighroller]

This board sounds as good as the 780i that I took back to the store within three days.

[|SiLA|]

Could you post your read-write-copy-latency results and voltage setting , please?

The reason why I ask is if you for example use P03 BIOS, performance of your memory is limited too much to satisfy you with nice timings, but it doesn't mean that you have good results.

Take a look on the following pictures (BIOS P03, no corruption issues):

[IM]http://www.audio-measurements.com/0/images/12.png[/IMG]

Settings almost as yours (1800 FSB vs yours 1700, but 7-6-6-15-1T vs yours 7-6-5-18-1T), P03, no corruption issues:

[IMG]http://www.audio-measurements.com/0/images/6.png[/IG]

And finally my current setting with BIOS 790P03R2:

[IMG]http://www.audio-measurements.com/0/images/13.png[/IG]

Timings don't look so nice, but the memory performance is significantly improved, and still no corruption issues.

i have the SIIE with 0512 bios

i'll post tonight

[|SiLA|]

here's the screens

copy




latency




read




write


i've tried to run unlinked mem at 1800 same timings but it's slower

[slim142]

Saaya
That rumor about C1, you are saying might be true? Remember when Anandtech reviewed the 790i they mentioned an Engineering Sample C1 which performed nicely?

Dont know what the heck is going on here but it might be better for me to hold on the credit card before purchasing anything.

For SIIE owners, have you tried BIOS 0601?

[GAR]

http://www.evga.com/forums/tm.asp?m=332217

P04 is out!!!

[gmcg]

do you guys use system ghost to recover from corruption?
I just don't feel like re-installing windows every time the slightest OC will corrupt it.
Any hint what tricks / software you use for testing corruption, restoring windows, etc.... ??

Image - Acronis. Tricks - scandisk - right click on your HDD, select "properties", select "Tools", select check HDD (upper button). It will run as soon as you will reboot.
Restoring windows - use your installation CD, there is an option in it. But image solution is easier and faster.

e_f, well then that settles it for sure, it MUST be the SPP...
what? that doesnt make any sense, why would you implement a mechanism to "check" if a certain speed is stable and only then allowing the system to boot? this doesnt really sound right... are you sure thats how other chipset mfgs do it?

That was my assumption, it could be wrong, of course.
Logically - it should be some kind of checksum during the system self-test. I don't know exactly how BIOS is programmed regarding algorithm of memory test or booting ability, may be someone knowledgeable enough can enlighten us?

well this sounds like first you release a bios thats tweaked for performance, so then all the reviews look great and some people post great results, but then you release a new bios that actually makes all boards stable but performance sucks... so basically they are selling 2 types of 790 ultra boards, some that can run stable with a tightly tweaked bios and some that cant... come on, wth is that? i honestly pray its not like that...

Well I guess it was a temporary action to stabilize the situation, people report that just released BIOS 04 works fine.

yeah, totally agree... but then again, some boards seem to work perfectly fine even with tweaked bios and high clocks...

if your system is within your stable limits, why not?

i keep hearing about different chipset revisions... b2 and c0?
so the older rev is actually better?
what revisions do you guys have?
c0 seems to be really crappy while b2 steppings seem to work fine?
are there c1 and b1 revisions too?

Need to get some more info about this issue.

[gmcg]

here's the screens

Please don't use the release version of Everest, use the beta 1347 or better 1352
Your read speed looks fine, latency as well, but write speed @7k+ is too low.

[|SiLA|]

weird i keep getting 7k only on write

[BenchZowner]

And also...running the Cache & Memory Benchmark is easier, and more convenient on screenshots ( as it includes all memory tests in a small window ) [ it's under the tools menu if I recall correctly ].

From tomorrow and on I will be hammering the Striker II Extreme, so I'll have my conclusions about this Data corruption issue soon.

[|SiLA|]

And also...running the Cache & Memory Benchmark is easier, and more convenient on screenshots ( as it includes all memory tests in a small window ) [ it's under the tools menu if I recall correctly ].

From tomorrow and on I will be hammering the Striker II Extreme, so I'll have my conclusions about this Data corruption issue soon.

hope to see from you soon any comments on 0601 bios

[slim142]

Report results w/ new P04 BIOS

[gmcg]

Report results w/ new P04 BIOS

Absolutely happy with 04, stable 7-6-5-18-1T@2.0V memory in linked sync mode with QX9650 CPU@4072MHz (temporarily with AC while I am waiting for a new case), 22464 3dmark06 with only one 9800GX2 (stock cooling).
http://service.futuremark.com/compare?3dm06=6348994

No corruption issues so far. Read-write-copy-latency results could be a bit better, but I will get them better with WC, I am sure.
Still "day and night" with P03 (low results) and even 790P03R2 (unstable, corruption issues).

Bravo, nVIDIA!