790 corruption issues?

[tekjunkie]

IDK if there is any P1 and P2 on the 790i reference boards, but if there is one enable it and give it a shot.

[manyga]

Very small pics bro!...

the same results that me and my q6600 with ddr2!

I hope better results with eVga 790i ultra and OCZ 2x1gb PC3-12800 CL7!

Regarddss from SpaiN!

[dinos22]

Check out the difference in the Everest memory scores between EVGA 790i and Asus P5E3 Premium. Same cpu on each board (q9450 at 8x450), same ram (2G Cellshocks at 1800, 8-7-7-20, all other settings on auto). Of course this is untweaked and the boards obviously use different auto defaults but still ...

790i:



X48:

lol the funniest thing about this shot is actual times

but once you get a bios with P1 and P2 and enabled them bandwidth will equal or exceed that X48 board straight from bios startup

but bandwidth is one thing.....efficiency another
they don't always go hand in hand linearly

[RealTelstar]

Check out the difference in the Everest memory scores between EVGA 790i and Asus P5E3 Premium. Same cpu on each board (q9450 at 8x450), same ram (2G Cellshocks at 1800, 8-7-7-20, all other settings on auto).

I CANT read the everest numbers with such tiny and compressed pictures
Did you optimize the latencies of the x48 board?

[Aivas47a]

Sorry about the small pics guys. I'll try to edit and replace with bigger pics after work.

Basically what the pics show is much slower default memory read/write (like several thousand points) and higher latency on the 790 vs. the X48. Cache performance numbers are almost the same.

The numbers for both boards are strictly using default/auto memory settings except for manually setting 1800 and 8-7-7-20 on the main timings.

Everest performance for both boards can be improved a lot with tweaking, I just found it interesting that the "plug and play," default performance of the boards would show such a big difference.

The difference in memory bandwidth numbers does NOT translate into any noticeable difference in super pi (the 790 actually has the better 1M time) or 3DMark06 (results using same video card are not in my post but are almost identical).

dinos/others, forgive my ignorance but what is the P1/P2 setting and how should that be tweaked?

[RLM]

Had some corruption, but have opted for running memtests before booting up, which has helped.

Apart from that very dissappointed. Doesn't like my ram at all. Won't even get through bios in anything other than 1:1. Have tried v.loose timings. Will run small FTT no probs, blend fails in seconds.

Have tried P03 and P03R2. P03, which was downloaded from the EVGA forum, would only flash if forced.

Have spent a good 24 hours solidly testing, and basically

Is it a bios issue? Is it hardware? Is it ram compatibility? Should I sell the board? Buy new ram instead? A few questions coming to mind right now.

In answer I think unless a new bios crops up in a few days, I'll take the cheaper route, and flog the board for something else.

One positive. It looks nice.lol

RLM

[Aivas47a]

New review from FCG at Anandtech today -- he basically loves the 790i (Striker version), with the only caveat being the data corruption issue:

http://www.anandtech.com/mb/showdoc.aspx?i=3283

[aoch88]

Can anyone tell me what's good about this nvidia 780i compared to the P35 or X48?

[BenchZowner]

Can anyone tell me what's good about this nvidia 780i compared to the P35 or X48?

This is the 790i.
It's the first nVIDIA nFORCE chipset with DDR3 memory support ( no DDR2 support at all ).
It's a brand new chipset design, not like the 780i which is a newer revision of the 680i + a bridge chip ( PCIe 2.0 controller ).
It has far greater overclocking capabilities ( very high average FSBs ), and decent DDR3 overclocking capabilities.
And of course, it supports the 45nm Quads.

vs the P35/X38/X48...
It reaches higher FSBs ( in average ) with the E8xxx & Qxxxx CPUs than P35/X38/X48 [ some decent X38/X48 boards can come pretty close with the highly clocking 790i's though ].
It supports 2-way & 3-way SLI...while P35/X38/X48 support only CrossFire.

That's pretty much it.

P.S. Please lets keep the thread on topic now.

[Gylen]

Sorry about the small pics guys. I'll try to edit and replace with bigger pics after work.

Basically what the pics show is much slower default memory read/write (like several thousand points) and higher latency on the 790 vs. the X48. Cache performance numbers are almost the same.

The numbers for both boards are strictly using default/auto memory settings except for manually setting 1800 and 8-7-7-20 on the main timings.

Everest performance for both boards can be improved a lot with tweaking, I just found it interesting that the "plug and play," default performance of the boards would show such a big difference.

The difference in memory bandwidth numbers does NOT translate into any noticeable difference in super pi (the 790 actually has the better 1M time) or 3DMark06 (results using same video card are not in my post but are almost identical).

dinos/others, forgive my ignorance but what is the P1/P2 setting and how should that be tweaked?

That's weird. My DDR3-1333mhz OCZ 7-7-7-20 scores better than your DDR3-1800? (Using 790i board also) I get a score closer to 10k for read/write. There must've been something wrong with your 790i test?

[Aivas47a]

That's weird. My DDR3-1333mhz OCZ 7-7-7-20 scores better than your DDR3-1800? (Using 790i board also) I get a score closer to 10k for read/write. There must've been something wrong with your 790i test?

Probably so ... the numbers are so off I was wondering if there was some glitch with the 790 memory controller that might have some relationship to the data corruption issue, but no one else seems to be seeing this (including in FCG's new Striker II Extreme review at anandtech today) so probably my test run was just borked somehow. (I have had data corruption too a couple of different times when pushing high FSB -- something that can happen on any board but seems to happen more easily on this one.)

[Gylen]

I also noticed your FSB is higher than mine (450, mine is at 333) and I've read that it effects synthetic benchmarks such as this one. There was definately something wrong with your test. But like you said, it probably has to do with the memory issue the 790i is having.

[eternal_fantasy]

Probably so ... the numbers are so off I was wondering if there was some glitch with the 790 memory controller that might have some relationship to the data corruption issue, but no one else seems to be seeing this (including in FCG's new Striker II Extreme review at anandtech today) so probably my test run was just borked somehow. (I have had data corruption too a couple of different times when pushing high FSB -- something that can happen on any board but seems to happen more easily on this one.)

What version of BIOS are you using?

[Aivas47a]

What version of BIOS are you using?

P03 from the evga website.

[slim142]

Is Gigabyte and the others like DFI and MSI going to release 790i motherboards? if so, when? People need more choices imo... (well i know, with things actings like this, we dont really need more for now until a fix is released)

[BenchZowner]

Gigabyte is planning to release a 790i board, but they have no ETA yet.
And no early E.S. ready yet. [ official info ]

[dinos22]

Gigabyte is planning to release a 790i board, but they have no ETA yet.
And no early E.S. ready yet. [ official info ]

yup
they are not in a rush hehehe

[loonym]

Here's the P7N2 Diamond http://global.msi.com.tw/index.php?f...=1&cat2_no=170

[slim142]

yup
they are not in a rush hehehe

I can tell

[BenchZowner]

yup
they are not in a rush hehehe

Wise move for the time being I'd say.
I mean... would you release a product known to have severe issues ( data corruption ) ? I for one wouldn't... not until the solution or at least things clear out and people know what, how & why.

I can tell

Patience is a virtue...

-------
Norton Ghost installed...image taken...

Currently benching SuperPi 32M @ 500MHz FSB + DDR3-2000 ( Link + Sync Mode )

[BenchZowner]

Norton Ghost installed...image taken...

Currently benching SuperPi 32M @ 500MHz FSB + DDR3-2000 ( Link + Sync Mode )

Norton image was worth it
Restoring image now.... corruption #1 for today ( it's my second one for the record ).

Oh by the way...the 32M stopped at loop 24 ( LOL ) [ computer froze ].

[dinos22]

Wise move for the time being I'd say.
I mean... would you release a product known to have severe issues ( data corruption ) ? I for one wouldn't... not until the solution or at least things clear out and people know what, how & why.

well that wouldn't have been the reason for delay but this looming as well i am sure they will also keep an eye on that

actually i'll fire off an email to my contacts in case they are not aware

[slim142]

Wise move for the time being I'd say.
I mean... would you release a product known to have severe issues ( data corruption ) ? I for one wouldn't... not until the solution or at least things clear out and people know what, how & why.

Patience is a virtue...

Yes, it is a wise move. Not that Im desperate for more options/choices but is just that Im surprised others didnt release their boards on time. Thats normal for DFI but not for others...

[RealTelstar]

Just to be sure, anybody got OS or data corruption issues from a single drive or array connected to a third party controller (Areca, LSI, Dell etc)?

This should be a 100% no.

[BenchZowner]

Just to be sure, anybody got OS or data corruption issues from a single drive or array connected to a third party controller (Areca, LSI, Dell etc)?

This should be a 100% no.

Unfortunately it's a YES