Gigabyte EX58-UD5/Extreme Discussion Thread

[DavyBoy]

LoL.. Mate, you aren't listening, this temps I am talking about are at stock, voltages are like 0.954v when idling.. I'm not even talking about the temps when I o/c this chip, so you talking about me upping the voltages are completely irrelevant..

And it has nothing to do with bad mounting or thermal management like you have stated, I have tested with 3 different HSF's, and I have reapplied TIM on several occasions.. and the variation between cores is only around 3c of a difference as well...

Even with the same cooler I used on my last chip which would have it idling at 36c there was still a 10c difference between these chips..

Like I said, this matter is over, as it has been dealt with..

[jalyst]

if you wanted a chip with guaranteed overclocking potential - we offer a bundle based on that that i personally create and test for each customer

Remarkable, I never knew such retailers existed....
I wonder what sort of premium one must pay for a "guaranteed" (R) OC'er?

I don't think anything like that exists in Oz, now there's a market!
If you're prepared to deal with the fanatics!

-jed.

[jalyst]

LoL.. Mate, you aren't listening, this temps I am talking about are at stock, voltages are like 0.954v when idling.. I'm not even talking about the temps when I o/c this chip, so you talking about me upping the voltages are completely irrelevant..

And it has nothing to do with bad mounting or thermal management like you have stated, I have tested with 3 different HSF's, and I have reapplied TIM on several occasions.. and the variation between cores is only around 3c of a difference as well...

Even with the same cooler I used on my last chip which would have it idling at 36c there was still a 10c difference between these chips..

Like I said, this matter is over, as it has been dealt with..

Yeah please do drop it now mate,
The thread is getting too focussed towards your dramas with a retailer.

Cheers

[Hornet331]

The reason I asked, because when I tried to boot to Vista, I got a black screen(after the loading bar). However, when I booted to Safe Mode, it booted just fine. I just could not boot to Vista for some reason. I pressed the Clear CMOS by the I/O plate, and restarted my PC and it booted just fine. The time did not change at all though.
Do you think my CMOS battery is getting weaker???

When you can boot in savemode its usualy a driver that s up windows. You have two options:

A) Use a restore point, if you have one
B) Reinstall the OS

I would go for B.

[123456iambelial]

When you can boot in savemode its usualy a driver that s up windows. You have two options:

A) Use a restore point, if you have one
B) Reinstall the OS

I would go for B.

I cleared the CMOS and it works now. I am thinking the culprit is the sound card driver.

[aver]

Just wondering if someone could give me help on how to lock the multiplier in the bios. In windows the multiplier fluctuates between 15 - 21 depending on load.

[123456iambelial]

Just wondering if someone could give me help on how to lock the multiplier in the bios. In windows the multiplier fluctuates between 15 - 21 depending on load.

You have to disable all Intel's power saving. They are under Advance CPU Feature

[aver]

You have to disable all Intel's power saving. They are under Advance CPU Feature

Thank you very much. I see you run Vista 64 do you have issues running SuperPi 32M? The only version that can pass 32m on my machine is SuperPi 1.1e everything above crashes. Starting to think its either Vista 64 or I need to reinstall.

[123456iambelial]

When I ran SuperPI, it did not give me any kind of problem. Right now, I am trying to solve my problem with my sound card. It is a serious problem.

[DavyBoy]

Just wondering if someone could give me help on how to lock the multiplier in the bios. In windows the multiplier fluctuates between 15 - 21 depending on load.

Disabling C1e and EIST in Advanced CPU Features will prevent the multi from dropping.

I personally prefer to leave these settings enabled in the BIOS, as it saves energy, and as far as my experience goes, they don't have any effect on stability.

[aver]

Disabling C1e and EIST in Advanced CPU Features will prevent the multi from dropping.

I personally prefer to leave these settings enabled in the BIOS, as it saves energy, and as far as my experience goes, they don't have any effect on stability.

Ya I was just trying to see what kind of SuperPi Times i could get.. Got a 9.77 @4.2Ghz but the CPU started at 15x multi and moved to 21 after the start..

[M31]

Like I said, this matter is over, as it has been dealt with..

I hope to your satisfaction at last, m8.

The true test of a retailer is when a deal goes wrong ... thankfully my recent purchases from them seem to be OK so far, but if there does turn out to be a problem, then I foresee nightmares getting it sorted.

OcUK are famous for bad after market service, reneging on assurances and delays, my and friends experiences, as well as googeling OcUK customer service back this up.

This time ... I swear, I'll never use them again, even if I have to wait for other UK retailers to get stock in.

In fairness, Rich from OcUK seems like a decent guy, but he's bound by his employers ... I have no grudge against most OcUK staff, its those in charge that are at fault and why OcUK's reputation seems to be going downhill fast again.

[phsinc1]

anyone is using this ram "TR3X6G1600C7" with his UD5 or UD4P, I can get a stable 200x20 configuration but I must have all the memory sub timmings @ Auto, if I try to change for exmaple TRTP [auto set it to 9] for 6 like the SPD says, it will boot but give me error about TPM encryption press F10 to continue then it will recover from a bad OC and it won't boot again and I have to clear cmos, I tried with XMP profile and without XMP profile, with some seach I knew that this ram use BBSE Elpida chips, I'm using the latest beta bios.

[stasio]

Gigabyte Easy Tune6 B09.0818.1

[Lsdmeasap]

anyone is using this ram "TR3X6G1600C7" with his UD5 or UD4P, I can get a stable 200x20 configuration but I must have all the memory sub timings @ Auto, if I try to change for example TRTP [auto set it to 9] for 6 like the SPD says, it will boot but give me error about TPM encryption press F10 to continue then it will recover from a bad OC and it won't boot again and I have to clear cmos, I tried with XMP profile and without XMP profile, with some search I knew that this ram use BBSE Elpida chips, I'm using the latest beta bios.

Sorry, I do think I answered this for you at Tweaktown maybe not, but I was a bit hasty in my reply and forgot to tell you the rules concerning settings. Your mention of TPM Error screens made me feel I had to chime in and help you out with at least a reply, then it grew!

I spent a long time speaking with Gigabyte and the BIOS Engineers on the TPM error issue when setting advanced memory timings, so I thought I would let you know about all the rules and my findings. The error you are seeing I do believe is not caused by tRTP, and you can verify that by setting everything to auto and then only trying to set tRTP.

I do believe you may be seeing this error due to other incorrect settings, please spend a few minutes and take in the information below. Hope this helps you out, I know I was frustrated for a long time in regards to TPM Errors and memory timings. I think it is all resolved now though

You must follow this rule when setting tWTR:

tWTR Must be Write to Read Delay/Same Rank - (tWL + 4)


Also because these settings can cause the same or similar TPM errors for you if set improperly these rules must also be followed:

tWTP Must = tWR + tWL + 4
tWL Must be CAS Latency -1


** Sometimes I find tWL does not correspond correctly in regards to the rule and may be +1 what it should be, so keep that in mind also when setting either tWTP or tWTR. So find and set this, and test with Memset to find the actual Value in the BIOS then apply to the above rules. For instance I found at CAS 7 I can only set tWL to CAS Latency + 1, around 1600Mhz that is. And at CAS 8 I could either set tWL to Match CAS Latency value or CAS latency +1 again.

The above rule concerning tWL should be CAS Latency - 1 was given to me by Felix. So I do believe it is correct, but maybe not for X58 anyway? Not sure if he tested this or if it actually applies to X58 as I could never get it to apply properly following those rules. I have not spoken with him in a few months about it though, maybe he has done more testing or reading and could better answer you now on tWL and X58 as to why this happens.

And there is an offset between what you set in the BIOS for Write to Read Delay/Same Rank and what you are shown in windows/memset, this varies from different main timings, multiplier and freq, so you will need to test setting at a given speed until you know what the offset is. Then you can adjust the rest or not accordingly to change this value. I also believe this is directly affected by tWTP and thus also it's related settings. More thoughts by Eva2000 on this are linked below

Additionally the above statement on tWL Rules, must follow the intended tWL Rule (CAS Latency -1) and not the BIOS settings for tWL.

** This statement may be disregarded after some quick testing with F9c (UD4P) for a reply to this post. I will leave it for others just in case it may apply to their board or if they are using an older BIOS. It seems after a quick test to reply to you, that this issue I found with previous BIOS in regards to the above tWL rule may have been corrected now. I can set tWL to CAS latency - 1 now, so looks to now be resolved.


The Write to Read Delay / Same Rank setting is in the turnaround settings in the advanced memory timings page, per each channel

Eva2000 also posted some of his strange finding for a few settings in relation to DFI Boards. I am not sure if his findings are the same as mine, or if we both have different thoughts or therories on the matter.
http://i4memory.com/f80/initial-note...8-t3eh8-18821/

I was doing some tWTR and tWTP Testing but got side tracked and have not ventured further into things yet, but I have fully tested a wide range of tWTP and had an ongoing discussion with Gigabyte several months ago on the subject of memory timings and getting the same errors you mention when setting certain ones. After some work and several Beta BIOS's to try and resolve the matter I finally got them to add in tWL and tWTP to try and alleviate the issues mentioned. I just have not had time lately to test things in regards to this matter and so our discussions with the BIOS engineers and testing have not continued since late march.

I have ran a quick test for you just now, hope this can better show you with the above rules how things apply.

Here is a screenshot of Memset with the settings below applied. BIOS F9c UD4P

tWTR ...................................... 2 (Must be Write to Read Delay/Same Rank - (tWL + 4) So here we have Write to Read Delay/Same Rank 12 - tWL 6 + 4 = 2
tWR ........................................ 10
tWTP ....................................... 20 (tWTP Must = tWR + tWL + 4) So here we have tWTP 20 = tWR 10 + tWL 6 + 4
(Grrr After all the hassle and testing I went through on this setting, today I find a variance of at least 2 is allowed in the current BIOS I am using!!! Anyway, stick to the rule until later, then you can fiddle )

tWL ........................................ 6 (tWL Must be CAS Latency -1) So here we have CAS Latency 7 - 1 = 6
tRTP ....................................... 6





And the BIOS Settings I used for this quick test to show you how the above info works when setting these.






Please note, these settings may or may not be optimal (likely not) I just set these as a quick test to verify that the TPM boot issue was not present and to try and show how these settings and rules apply to all things considered.


Sorry for all the confusion this post may cause but I had to reply since this had bothered me for quite some time. I spent several months and MANY Emails back and forth trying to work with Gigabyte and the BIOS Engineers to try to get these settings corrected and added to the BIOS. And with our discussions on the subject stopping back in March due to my lack of testing time and never getting back with them on the subject I was not sure if they had fixed things yet or not, and which of my previous findings still applied.

Hope this helps everyone, and not makes things worse

@ Eva2000

Care to chime in with your additional thoughts or findings if you are reading this? It would be greatly appreciated!!

[phsinc1]

Thanks LSD, your settings worked fine and CPUTweaker showed correct values.

What concern me now is the "Round Trip Latency" I noticed that this value keep changing every time I reboot, sometimes it will be 51/51/54 and sometimes it will be 50/52/55, it is better to set it manually?

* I noticed in your screen, RTL is reporting 56 in bios and 55 in CPUTweaker, so I think your RTL is changing automatically too everytime

[orig3n]

Disabling C1e and EIST in Advanced CPU Features will prevent the multi from dropping.

I personally prefer to leave these settings enabled in the BIOS, as it saves energy, and as far as my experience goes, they don't have any effect on stability.

It can have a big effect as it may throttle back the CPU and undervolt it too much causing it to crash out as you will still have a high BCLK. Ive always disabled it on big overclocks since the Core2duo's to keep everything level and stable. It will also play with your idle temps as the voltage is pulled from the chip and reapplied dynamically.

[esdxx]

Gigabyte Easy Tune6 B09.0818.1

It gives me error during installation. Anyone else tried this?

[DavyBoy]

It can have a big effect as it may throttle back the CPU and undervolt it too much causing it to crash out as you will still have a high BCLK. Ive always disabled it on big overclocks since the Core2duo's to keep everything level and stable. It will also play with your idle temps as the voltage is pulled from the chip and reapplied dynamically.

The only time I have noticed these settings alter my voltages on the UD5 is when I have Vcore set to auto, but when I manually set Vcore and have these options enabled, the only thing that it appears to alter is the CPU multiplier.

Though I am only going by software that monitors voltages, they could very well be changing, and just the software doesn't pick it up.

But as I said, when I put everything back to optimized defaults in the BIOS, when I check CPUZ my Vcore is around 0.9445v idle, and they shoot up to 1.1675v when I put a load on it.

I have tested my o/c with these settings enabled and disabled, and I haven't witnessed any ill effects.

[Lsdmeasap]

Thanks LSD, your settings worked fine and CPUTweaker showed correct values.

What concern me now is the "Round Trip Latency" I noticed that this value keep changing every time I reboot, sometimes it will be 51/51/54 and sometimes it will be 50/52/55, it is better to set it manually?

* I noticed in your screen, RTL is reporting 56 in bios and 55 in CPUTweaker, so I think your RTL is changing automatically too everytime

No, those settings are not optimal like I said in my post, I just set them like that to show you that they all do indeed work. You will need to find your own stable settings, so please do not use those as they are not stable or optimal, only intended to show you the settings all work.

What you see in the BIOS on the left can and should be ignored. Those are almost always (99%) incorrect and not used. Of course sometimes a few are correct but you should just get yourself in the habit of not even paying attention to them to be on the safe side.

RTL can be affected by many things, QPI/Vtt voltages and QPI Speeds, main timings, and other timings as well.

I made a small note on setting them manually here, but please do read Eva2000's tips as well also mentioned in the above linked thread
http://www.xtremesystems.org/forums/...postcount=3107

It gives me error during installation. Anyone else tried this?

Did you uninstall your older one? You must uninstall the older version before you install the new one. So just checking

[DavyBoy]

I was wondering if anyone could tell me why EasyTune6 doesn't monitor my +12v and 3.3v?

I only see Vcore, +5v, and VDimm, but I have seen other peoples screenshots where it shows +12v and 3.3v values as well.

[MonsterDK]

Question about vcore & Vtt...

If I easily pass LinX over 20 passes and Prime 95 on low and high fft's.
I've found the sweet spot for Vcore. The next notch down fails LinX. Should I bump the sweet spot up 1 notch to ensure stability or assume solid stability from testing?
Vtt, I started at 1.355 and I'm down now at around 1.2*. How should I know where the sweet spot is here and should I apply the same principle of, once you sweet spot is found add 1 notch?

[Lsdmeasap]

I was wondering if anyone could tell me why EasyTune6 doesn't monitor my +12v and 3.3v?

I only see Vcore, +5v, and VDimm, but I have seen other peoples screenshots where it shows +12v and 3.3v values as well.

Different models have different sensors, I only see Vdimm and +5V on the latest easytune, not sure about in the past though? So it could be easytune errors, not sure

I do know everest shows vcore of course, and Vbatt, Vdimm, and +5V on the UD4P. No 12V and this has always been a constant. I do think some boards monitor the 12v and others do not. It should match up with what you see in the BIOS Health section, so in my case I do believe Easytune has a big now that I say that because Vcore is not shown in the monitor tab as in the BIOS.

I will report this to Gigabyte and see what they have to say

[MonsterDK]

Help please!

OK I was advised that this might add problems that need working round but I had the opportunity to purchase another 6gb cheap new so I added another 6gb to my rig.

I now have the dreaded problem on post where it won't boot and eventually resets itself.

I have 1600 OCZ ram (12gb)
QPI Link: x36 (7.2Ghz)
Uncore: x 16 (3200Mhz)
Clocks are set as recommended (worked fine with 6gb)..
8
8
8
24
1

QPI/VTT I've put back to 1.355V for the moment.
IOH Core: 1.160v
Dram: 1.660v
CPU PLL: 1.800
QPI PLL: 1.200
ICH I/O: 1.500
ICH Core: 1.100v
Dram term: 0.750v
All other ch-a ch-b ch-c etc etc left on auto.

CPU's at 4.2 (1.35V) and was stable via LinX and Prime on 6gb.

Any ideas please what I need to change to get 12gb working ?

(I've just about got my head around things and now i'm baffled again)

I just tried dram at auto and it booted... hmm... wonder if I should try one notch above 1.660?

OK that never worked.

Need help lads please.

[MonsterDK]

Anyone??