UnOfficial EVGA X58 Classified Owners Thread

[dejanh]

Sorry, I was looking at the QPI link, I'll check again.

Since I checked before I've changed the memory from Mushkin at 2:8 to Corsair at 2:10. Now the computer won't boot past 2X plus 1. If I try 22 it just keeps trying to reboot. Could be that I need more VTT but I'm high enough already there.

Hey wax. Thanks for clearing that up. I found it very strange that Uncore would be down-clocking on its own. Anyway, now that we cleared up that issue we can get back to the now pretty clear Uncore problem. You are now another confirmation that Uncore with a D0 will only go 2x + 1 and no more in reference to DRAM speed.

What is going on here?

[stephen23]

Sorry, I was looking at the QPI link, I'll check again.

Since I checked before I've changed the memory from Mushkin at 2:8 to Corsair at 2:10. Now the computer won't boot past 2X plus 1. If I try 22 it just keeps trying to reboot. Could be that I need more VTT but I'm high enough already there.

have noticed that your CPU temp was up to 90 celcius?
isn't some parts is not installed well?

[dejanh]

have noticed that your CPU temp was up to 90 celcius?
isn't some parts is not installed well?

Uncore stuff that him, I, and a few others are talking about has nothing to do with cooling. My cooling is perfectly fine but I get the same problem. 3oh6 also runs only sub-zero with the same problems.

[Olly_K]

I think he may be referring to my screenshot . .

[Vega]

Any one else here with Kingston HyperX 2000MHz memorys? I have major problem so get them run well. I have atleast manage to get them running 9-9-9-24 1T, 1900MHz ATM but want more...
My E-leet monitoring 1.5V QPI VTT so I guess that enough, right? Maybe it is better to reduce speed to 1800MHz with 8-8-8-20 timings instead..?

[dejanh]

Any one else here with Kingston HyperX 2000MHz memorys? I have major problem so get them run well. I have atleast manage to get them running 9-9-9-24 1T, 1900MHz ATM but want more...
My E-leet monitoring 1.5V QPI VTT so I guess that enough, right? Maybe it is better to reduce speed to 1800MHz with 8-8-8-20 timings instead..?

2GHz DRAM may require VTT as high as 1.65V.

[Bootsy]

Here are my Kinston HyperX @ DDR3-2100. vtt is set at 1.42 32m stable and memtest stable.

[Vega]

Very nice...whats the key to get them over 1900MHz then....?
I use more voltage than you but no succes any way!)
Any other settings I forgot maybe.

What is your settings for MCH Strap and QPI Freq. selection?

[Bootsy]

Very nice...whats the key to get them over 1900MHz then....?
I use more voltage than you but no succes any way!)
Any other settings I forgot maybe.

What is your settings for MCH Strap and QPI Freq. selection?

MCH strap is 2:10 QP freq I left at auto. This board handles memory really well. I can run the same speeds on my Rampage II but need 1.55 VTT. Have you tried lower volts? I can run DDR3 2000 @ 1.35 VTT with 1.60 on the ram.

I have 2 kits of these and the both do well over ddr-2000. I am try to get 12gb stable @ ddr3 2100

Will they run with the 1866 XMP profile?

[anzial]

I feel like D0s require less VTT, indeed they are less stable with high VTT.. On the other hand, they don't like the uncore more than 2:1 or 2:1+1 uncore/ram ratio

[cstkl1]

MCH strap is 2:10 QP freq I left at auto. This board handles memory really well. I can run the same speeds on my Rampage II but need 1.55 VTT. Have you tried lower volts? I can run DDR3 2000 @ 1.35 VTT with 1.60 on the ram.

I have 2 kits of these and the both do well over ddr-2000. I am try to get 12gb stable @ ddr3 2100

Will they run with the 1866 XMP profile?

hmm if ure kits is running 2k C8 at 1.6
then 2100 C8 is at 1.7v
and for 12gb ure going to need around 1.75-1.8v for 2100

again nobody with 4.4ghz 2100mhz prime95 large fft custom 5.4gb or blend custom 5.4gb 2 hours stable screenshot??

[LardArse]

Uncore stuff that him, I, and a few others are talking about has nothing to do with cooling. My cooling is perfectly fine but I get the same problem. 3oh6 also runs only sub-zero with the same problems.

Maybe it varies with cpu, 3540 normal aircooling only ~ 4.2G Uncore 23x matched with 2:10 Rams meaning Uncore x2 + 3,
cpu and uncore synched...



3540 normal aircooling only ~ 4.2G Uncore 23x matched with 2:8 Rams meaning Uncore x2 + 7,
cpu and uncore synched...



Under ln2 some clockin of uncores:

Rams @ 2:12 Uncore @ 4.9G, 27x -> x2 + 3

[dejanh]

Maybe it varies with cpu, 3540 normal aircooling only ~ 4.2G Uncore 23x matched with 2:10 Rams meaning Uncore x2 + 3,
cpu and uncore synched...

3540 normal aircooling only ~ 4.2G Uncore 23x matched with 2:8 Rams meaning Uncore x2 + 7,
cpu and uncore synched...

Under ln2 some clockin of uncores:

Rams @ 2:12 Uncore @ 4.9G, 27x -> x2 + 3

How about this theory...

Could it be that when we were thinking the Xeon chips had locked multipliers for Uncore and QPI that they were in fact exactly the same or very similar to the i7 920 D0 parts and when work was done to enable the Xeon multipliers to function correctly it actually fixed them where as the desktop version of D0 920 chips stayed broken.

Thoughts? Could you repeat your series of tests with a regular retail D0 920? There was one other person on XS reporting working Xeons yet everyone with a D0 920 is coming with the same story...

[LardArse]

How about this theory...

Could it be that when we were thinking the Xeon chips had locked multipliers for Uncore and QPI that they were in fact exactly the same or very similar to the i7 920 D0 parts and when work was done to enable the Xeon multipliers to function correctly it actually fixed them where as the desktop version of D0 920 chips stayed broken.

Thoughts? Could you repeat your series of tests with a regular retail D0 920? There was one other person on XS reporting working Xeons yet everyone with a D0 920 is coming with the same story...

I'll do so when I have it back right now the retail 920 D0 is in a demo rig for someone else's booth.

[dejanh]

I'll do so when I have it back right now the retail 920 D0 is in a demo rig for someone else's booth.

Sounds good. I really want to see what happens in your test results.

[tekjunkie]

I have the Xeons(W3540,W3520) and a 920 D0. Maybe I could be of some help here.

[dejanh]

I have the Xeons(W3540,W3520) and a 920 D0. Maybe I could be of some help here.

Knock yourself out Run the tests with each of the chips and see what results you get. The test is to be able to clock Uncore past 2x + 1 or 2x + 2 of DRAM speed on the Classified board. For example, if the DRAM is clocked to 1600MHz, see if you can get Uncore of 3600MHz or higher, or if DRAM is at 2000MHz if you can get Uncore 4400MHz or higher.

[bluehaze]

again nobody with 4.4ghz 2100mhz prime95 large fft custom 5.4gb or blend custom 5.4gb 2 hours stable screenshot??

Would love to see this, My ram is stable at 2100, my processor also stable at 4.4 but combine the 2 and as soon as I start Prime the computer powers off. No errors or anything it just powers down and theres no Windows error when loading back up either.

[Vega]

I have a strange thing with my board and wonder if is normal or not?

I can easily multi clock my 965 from stock 133bclk...past 4.2GHz with out any problem with multi x32 but when i try higher bclk as 190 it wont boot at all with mult x16.
I did a test....cleared cmos and set every voltage to auto exept Vcore and Vdimm and my max bclk was ~180...is this normal, maybe a bios thing??
I know my 965 handle 210+ bclk on R2E.

Thanks in advance.

[LardArse]

I have a strange thing with my board and wonder if is normal or not?

I can easily multi clock my 965 from stock 133bclk...past 4.2GHz with out any problem with multi x32 but when i try higher bclk as 190 it wont boot at all with mult x16.
I did a test....cleared cmos and set every voltage to auto exept Vcore and Vdimm and my max bclk was ~180...is this normal, maybe a bios thing??
I know my 965 handle 210+ bclk on R2E.

Thanks in advance.

there are 3 things you need to set,
VTT ~ +200mv
DRAM ~ 1.65v
QPI Link Speed @ 4.8GT/S

[FragMagnet]

Here's what I got with my air cooled W3520 on the 760. The vtt was stock at this speed until I upped the uncore. I found stability at 17x with +.05mv, and 18x with +.1mv I was using IBT 2x 10 passes to test the settings out before launching into prime95 blend, it hiccupped on the 2nd IBT, so I upped the vcor 1 notch and the vtt +.025mv just to be sure. 19x wouldn't boot at this setting using 2:8 on the mem.
Remember e-leet seems to show .09v higher than set and measured with a DMM, the rest are very accurate

[Waxking1]

Hey wax. Thanks for clearing that up. I found it very strange that Uncore would be down-clocking on its own. Anyway, now that we cleared up that issue we can get back to the now pretty clear Uncore problem. You are now another confirmation that Uncore with a D0 will only go 2x + 1 and no more in reference to DRAM speed.

What is going on here?

I don't think its cleared up in as 2x plus 1. I still booted at 2x plus 3 with my Mushkin @ 1600, I just was looking at QPI Link in Eleet. The problem with the Corsair not booting a higher uncore than 2x plus 1 is most likely voltage.

[jasonelmore]

nice thread. See if you can get as organized as my Rampage II Extreme Thread

[Vega]

there are 3 things you need to set,
VTT ~ +200mv
DRAM ~ 1.65v
QPI Link Speed @ 4.8GT/S

I tryed but it did not work....no boot at 190x16.
I used mem. mult x8 so the memorys cant stop it.

Any other ideas?

[dejanh]

nice thread. See if you can get as organized as my Rampage II Extreme Thread

I doubt it will get as nice as that thread. That's a very impressive thread

Here's what I got with my air cooled W3520 on the 760. The vtt was stock at this speed until I upped the uncore. I found stability at 17x with +.05mv, and 18x with +.1mv I was using IBT 2x 10 passes to test the settings out before launching into prime95 blend, it hiccupped on the 2nd IBT, so I upped the vcor 1 notch and the vtt +.025mv just to be sure. 19x wouldn't boot at this setting using 2:8 on the mem.
Remember e-leet seems to show .09v higher than set and measured with a DMM, the rest are very accurate

This is a Xeon and they do seem to work fine. Nobody with a Xeon reported this kind of behavior yet, not that I registered anyway. Also, your DRAM is running below stock settings (just over 1200MHz vs. rated 1600MHz) so I am not sure if that has any impact on anything.

I don't think its cleared up in as 2x plus 1. I still booted at 2x plus 3 with my Mushkin @ 1600, I just was looking at QPI Link in Eleet. The problem with the Corsair not booting a higher uncore than 2x plus 1 is most likely voltage.

Both of you guys seems to be alluding to voltages. I can assure you that they are not a problem. I have tried all voltage combinations with no success. Raising VTT or DRAM voltages makes no difference. Unlike with your chips my D0 does not even POST if Uncore is more than 2x + 1 in reference to DRAM speed. It stays stuck looping F3/68 codes on the ECP. However, like I said before clocking the DRAM higher allows me to clock the Uncore higher as well, as long as I stay within 2x + 1.