Official GIGABYTE Forum
Questions about GIGABYTE products => Motherboards with AMD processors => Topic started by: DrStrange on August 23, 2009, 10:45:58 pm
-
Hi there,
I just put together a computer including an MA770T-UD3P, a pair of Crucial 2GB DDR3-1333 sticks (part no CT2KIT25664BA1339) and a Phenom II X2 550 Black Edition. When I look at the PC Health Status page the vdimm is 1.616v. Everest is saying that the vdimm is 1.62v. I am using the latest F3 bios and all values are on auto. The DRAM voltage control in the BIOS starts only from 1.600v, but the manual says it should start from 1.500v.
Can somebody please tell me how do I get the vdimm back down to 1.500v as I want to run at stock?
-
Hi,
Tech support may be your best option
http://www.gigabyte.com/support-downloads/technical-support.aspx
regards
-
Also check with Crucial on the memory's requirements. Some DIMMs are designed to run at higher voltages (my own Corsair 1600MHz sticks need 1.8V at that speed or 1.6V at 1333MHz). I think some sticks talk back to the board and set the minimum voltage to suit their requirements. It is quite possible your sticks want a minimum of 1.6V and the BIOS is ensuring it gets the voltage it wants.
-
I'll add to this thread as I have a similiar problem.
Same motherboard, bios F2, unlocked Phenom II x3 720BE, 2x2GB OCZ Intel XMP 1600 cl8 ( funny, I know ). Now, my problem, even with previous RAM ( 2GB A-Data 1066 cl7 ), is that bios and all monitoring programs report dimm voltage ~0.1v higher than set in bios. I have no idea why as other voltages are about accurate. I THINK that the reported voltage is wrong as RAM didn't boot with supposedly accurate voltage ( set to like 1.520-1.540, reported ~1.650 ). Any ideas why it's showing values this much higher? Right now it's set to 1.660 and reported as 1.780 :o Faulty voltage sensor?
edit:
Right, forgot to mention, it's the first version of the board ( just in case ).
-
Hi there,
Sometimes these kind of issues are dealt with by an update to BIOS.
The latest version for your Mobo is F11C so it might be that the voltage reporting issues has been addressed in one of the updates released since the F2. Check here: http://www.gigabyte.com/products/product-page.aspx?pid=3096&dl=1#bios
I would suggest that you try flashing BIOS with a newer version. If you are not sure how to do that check out the FAQ here: http://forum.giga-byte.co.uk/index.php/topic,2441.0.html
As you are only at F2 I would not recommend jumping straight to the F11C BIOS but would advise that you update your BIOS in several small steps F2 - F5 then F5 - F8 then F8 - F10 this is not essential but is a safer method.
-
Ok, I've updated bios to F10 but except for getting a few additional options and losing quite a few ( like about half of voltage settings ) there's no improvement.
I'm like 99% sure now that the reported values are wrong, tho. I've been playing with oc'ing a bit and seems like RAM isn't fully stable below 1.680v ( technically, it should work with 1.65v ). That's 1.790v reported which, if it were real, would probably actually decrease stability in this case. Any further ideas?
-
Did you clear the CMOS as well as updating the BIOS? If so then I would think it's a reporting error as you are running a new BIOS now. Maybe the sender is faulty.
-
Ok, I'm officially lost here. I haven't cleared CMOS, I may try that tomorrow. Anyway, turned out that RAM at 1.680v wasn't stable either. Rather, I couldn't get it stable with "stock" timings and speed at all trying everything from 1.520 to 1.700. Now, with set 1.660v ( so pretty much stock for these sticks ) and with multiplier one lower I'm running them at 1533Mhz 7-7-6-22, tRC-29, rest is the same as bios values were for 1600Mhz. I'm not 100% positive they're stable, but quick test runs through IntelBurn Test and MemTest didn't show any problems.
As much as I would love to believe that it's running these values with that voltage it seems kinda far fetched. But then, that would mean they're running at 1.780v ( once again, reported voltage )? I have no idea why it didn't want to work at 1600Mhz at all, either.
For the record, it should have worked at 1600Mhz 8-8-8-24, tRC-36 at 1.650v. That's the XMP profile for Intel processors so manually setting it to these values should work >:(. My main problem seems to be figuring out the REAL voltage my DDR3 sticks are getting.
-
I have no idea why it didn't want to work at 1600Mhz at all, either.
For the record, it should have worked at 1600Mhz 8-8-8-24, tRC-36 at 1.650v. That's the XMP profile for Intel processors so manually setting it to these values should work >:(. My main problem seems to be figuring out the REAL voltage my DDR3 sticks are getting.
Well AMD doesn't recommend running the memory at anything over 1333 because of risk of damage to the memory controllers that are now incorporated into the CPU. Not only that but 1600 isn't a supported speed anyway.
The XMP profile for Intel chips wouldn't help on an AMD platform anyway.
Please see this thread for information: http://forum.giga-byte.co.uk/index.php/topic,2515.0.html
-
Huh, I've skimmed over that article and read a bit around the net earlier but it's more ominous than I thought. One mystery solved then. Funny thing it's running at 1533Mhz without problems ( so far ). Maybe I should drop it down to ~1333Mhz though.
Anyway, I think I'll check if it'll work at 1333Mhz with adjusted timings with correct "reported" voltage. Who knows, maybe the board's really running my memory at ~0.1v higher than set.
Yeah, I know that XMP profiles won't work with AMD platforms but if you set everything as it is in the profile it, generally, should work.
-
Huh, I've skimmed over that article and read a bit around the net earlier but it's more ominous than I thought. One mystery solved then. Funny thing it's running at 1533Mhz without problems ( so far ). Maybe I should drop it down to ~1333Mhz though.
Yes, sorry I should have noticed earlier but as absic would be quick to point out old age is creeping up(well maybe not creeping). If you read absics report on the memory problem you will understand that even if you suceed in getting the memory to run at a faster speed you run the risk of burning out the CPU somewhere down the line. It even happened to one of AMDs engineers!
Yeah, I know that XMP profiles won't work with AMD platforms but if you set everything as it is in the profile it, generally, should work.
You would think so wouldn't you but often it doesn't work like that. In the same way that memory designed for Intel chips often doesn't work on AMD platforms and vice versa.
-
Hi there,
one of the main reason AMD do not recommend running RAM above 1333 Mhz is for this very reason, they cannot guarantee system stability. There is also the increased risk of damaging the Memory controller that is now built into the CPU. It is also a known issue with AMD, of systems seemingly running OK at with RAM at 1600MHz only for them to fall over for no apparent reason. This doesn't mean that you can't get your RAM running at 1600 MHz only that doing so increases the risk of system instability and/or killing the CPU.
Lecture over.
I had all kinds of challenges in getting my RAM to run at 1600 MHz before I learnt of the CPU limitations but I did get it to run. My RAM is Corsair Dominator CMD4GX3M2B1600C8 which has specified ratings of 1.65V and timings of 8,8,8,24. Setting these figures in BIOS didn't work and the system was falling over whenever I opened a web browser or any application. I backed the timings off a little to 11,11,11,29 and the system is stable. However, after learning of the problems with the Memory controller on the CPU I have reverted to 1333MHz @ 1.5V.
Still looking into the voltage reading issue and will post back regarding that when I've done a bit more digging.
-
Ok, I've cleared CMOS and it didn't help but I've figured the situation out ( or so it seems ). The problem is more serious than I thought as the voltage reported likely is the real one. In other words, min. voltage I can set is 1.612v. Also, good thing I didn't leave it with previous settings as I was cooking the ddr3 sticks. I'm sure OCZ didn't mean for them to run at 1.780v ::).
Anyway, why do I think it's like this? I'm running these sticks at 1333Mhz 7-7-7-22 right now @ reported 1.648v. Haven't testem them yet but I don't think they'd even boot at 1.540v.
Seems like a hardware issue somewhere ???. I'm not sure I want to try and RMA it as I've basically just finished putting this computer properly together. Well, I still have about 2 years of warranty on it so it's not like it can't wait. On the other hand, CPUID HWMonitor once showed a slight drop in voltage from 1.780 to 1.760 for a moment. Not sure if that kind of info is reliable, tho.
Right, as this may not be all that clear, I don't have the same problem as DrStrange. I can "set" voltage to 1.500 but in reality it's, if I remember right, 1.612. Generally, it gives it ~0.1v more than it should.
-
Hi,
I'm not running your board but I have now checked with my own, the GA-790XTA-UD4 and also with a GA-770TA-UD3 and both of these boards are reporting the voltages accurately. Of course, there is a slight variation in the voltage draw depending on what the PC is doing, with it dropping the voltage below the 1.65 I have set or going above that when I increase the load, according to EasyTune6, AMD overdrive and HW Monitor, which is normal.
The problem with getting the RAM to run properly though, could be due more to the Memory controller on the CPU rather than the voltages.
-
The RAM IS running properly, just the voltages are different than set. Unless you mean that this may be caused by the memory controller on the CPU?
I'm now certain that the reported voltages are, if not 100% correct, at least more like the real thing. At 1333Mhz I could get these sticks to run @ reported 1.7v as 6-6-5-18 and I'm still tweaking them a bit more. I'm quite happy with these values though ;D. Pretty much in the range of the early Reaper and ReaperX sticks at comparable voltage levels.
-
OCZ has always run their memory at higher voltages then most other companies.
-
That'd seem to be true as these sticks wanted 1.73v to run through BurnTest @ 6-6-6-19 1333Mhz. 6-6-5 is probably the limit if you're willing to up the voltage even more.
Anyway, I just stumbled upon some interesting thread:
http://www.overclock.net/amd-memory/610262-lowering-ram-voltage-gigabyte-ga-ma770t.html
"Hey Force, the GA-MA770T-UD3P will not go below 1.6V. In earlier bios revisions 1.5V was an option but the board over-volted to 1.6V. I think it's a hardware limitation of the board since the newest bios, F4, has a minimum Vram of 1.6. It will not hurt your RAM to run at 1.6V, and you may be able to tighten up your timings with the slight voltage increase."
Granted, that's just some random person who wrote this but the problem seems to be more widespread than just one or 2 boards. Seems like I shouldn't have the voltage settings below 1.6v availble in bios, though. Anyway, seems like a "hey it's not a bug, it's a feature!" thing.
Maybe I should just write an email to Gigabyte and ask them what's the min. vdimm for this board?
-
Maybe I should just write an email to Gigabyte and ask them what's the min. vdimm for this board?
That might be the best solution as the only reference I can find seems to show that 1.5V is the default for all versions of this board.
-
I know it is a different situation but on my last build(Intel platform) I used OCZ modules and had to run it at 1.95 to get it to run stable.
-
You know, next time I think I'll write a ticket pretending to be from the States or something ::). I think I'll be lucky if I'll get an answer before this week ends. Status after today still says it hasn't been touched by anyone.
Anyway, these memory controllers on Phenom II's are rather weak. I wonder what's the relation between them and CPU cores as I've noticed that the few times I tried upping voltage and raising CPU-NB beyond 2350Mhz ( last stable value, stock voltage ) it either wouldn't get stable no matter what or one of the cores got turned off. Anyway, I don't see it going much higher either way. Seems like it's near it's limits with stock settings. Guess that's only more of a reason to stay with RAM within 1333Mhz limit.
-
Sorry but I think I've got lost a little here, could you just confirm that you have unlocked an X3, that you are now trying to overclock the unlocked CPU and are wondering why you can't get the system stable?
-
I can certainly confirm that I've unlocked an X3 and am running it at 3.5Ghz ( when there's need, K10Stat ). The system is perfectly stable, though. I've just observed that CPU-NB on my processor is pretty weak as it doesn't really want to go beyond 2350Mhz. Just to keep the FSB stock I'm running it at 2200Mhz. I was just wondering if it may be limited further by that unlocked core. As I said, I noticed that when I up the voltage on CPU-NB it often shuts down that 4th core. I'm guessing some safety feature of the bios as it boots normally but the processor reverts back to 3 cores.
So, for the record, the system's indeed overclocked but stable. Actually, now I'm thinking about downclocking the graphic card for idle mode ;D
-
I am sure that absic will have plenty to say on this matter but for my own twopenneth if you are using one of the processors with an extra core unlocked that is a bonus anyway and the previously locked core might be flawed anyway and that is the reason for it being binned as a dual core or three core etc.
-
Pretty impressive O/C with the unlocked core. :o
Yes, I have to say that you are probably hitting the threshold with the CPU which is why it keeps falling over when you try and push it a little further.
Just a thought and, as I don't run an unlocked CPU I can't check this out, but in your BIOS is there the Core Performance Boost option? If so try disabling it and then see if you can gain a bit more from the CPU. The Core Boost feature is designed to shut down 1 or 2 of the cores on the CPU and boost the frequency of the remaining cores to improve system performance. Also, have you disabled the K8 Cool&Quiet feature in BIOS, this could be having an effect and holding you back too.
-
I am sure that absic will have plenty to say on this matter but for my own twopenneth if you are using one of the processors with an extra core unlocked that is a bonus anyway and the previously locked core might be flawed anyway and that is the reason for it being binned as a dual core or three core etc.
Yeah, I know. I'm rather happy that it's running stable unlocked. All in all, I've bought this PC used and didn't really know all that much about what'd I'd be getting. Considering the price, the fact that I also got receipts that mean I have warranty for 2+ years on everything I'm, indeed, quite happy. The processor aside, I've been positively surprised with the graphic card, too ( better cooling than stock and slightly factory overclocked ).
Pretty impressive O/C with the unlocked core. :o
Yes, I have to say that you are probably hitting the threshold with the CPU which is why it keeps falling over when you try and push it a little further.
Nah, it's not that great ;). It's a rather nice processor for overclocking:
http://www.overclock.net/amd-cpus/475649-amd-phenom-ii-x3-700-series.html
Some of these scores are probably fake, though.
Seems like you're right about the CPU. I've tried upping the CPU-NB to 2400 without voltage change and it booted without that last core again. Got worse as it didn't boot again at all after that. Had to clear CMOS to get it to run so I'm not playing with memory controllers any more.
Just a thought and, as I don't run an unlocked CPU I can't check this out, but in your BIOS is there the Core Performance Boost option? If so try disabling it and then see if you can gain a bit more from the CPU. The Core Boost feature is designed to shut down 1 or 2 of the cores on the CPU and boost the frequency of the remaining cores to improve system performance. Also, have you disabled the K8 Cool&Quiet feature in BIOS, this could be having an effect and holding you back too.
Nope, I only have that control that allows you to set % corrections to each core and the one that allows you to disable some of the cores ( or that's how I understand it ). K8 Cool&Quiet is disabled for this reason and so that it won't clash with K10Stat ( just a precaution ).
I think I've reached the limit of what's reasonable with this setup. I have 4 P-states for the processor:
- 800Mhz @ 0.7625v
- 1400Mhz @ 0.8875v
- 2800Mhz ( stock speed ) @ 1.2000v
- 3500Mhz @ 1.4125v
with CPU-NB bumped to 2200Mhz. To get anything more out of it I'd have to add loads of voltage so there's not much of a point. As I don't want to exceed 1.75v on RAM I'll have to accept the 6-6-6 timings. The sticks sit right next to each other if you want dual channel on this board and they're under the CPU fan ( Mugen 2 is huge :o ) so cooling isn't all that great.
Offtopic galore but do you have any idea if running the processor below it's rated operating voltage may damage anything? I've checked the specs on AMD site out of curiosity and noticed that my 800Mhz setting is running under their values.
I'm also wondering if downclocking memory on a graphic card could be harmful. I mean, it'd get more energy than it needs then ( as you can't lower the voltage )?
Back on topic: Gigabyte answered faster than I thought it would but for the moment it's a Q&A on my side rather than theirs. It'll probably take a while before they get all the info they want out of me.
-
Offtopic galore but do you have any idea if running the processor below it's rated operating voltage may damage anything? I've checked the specs on AMD site out of curiosity and noticed that my 800Mhz setting is running under their values.
Running the CPU under rated voltage won't cause any harm to the CPU. In fact, quite the opposite and I tend to undervolt mine as a matter of course. The only problem here is if you pull the voltage back too much, then the system, just like when O/C'ing, is liable to become unstable.
Never tried down clocking the RAM on a GPU so don't know the answer there but I would have thought that the graphic card would only draw the voltage it needed. There are never any warnings when you underclock only when you overclock! ;)
-
Update:
Ok, for the moment Gigabyte's answer is that they'll look into this and fix the bios if they'll find any problems. I suppose that's all I could ask for :)
-
Ok, I give up. I'm not even going to try to get anything more out of tech support. It's like playing Chinese whispers with them. All the meaning gets lost somewhere along the way.
They asked me to write to them after a week ( they were going to check this issue ). After waiting a few days for a reply I got an answer that's more or less, like this:
We put 2 sticks in, checked with CPUID and it showed 1.5v, then checked with PC Health Status which showed 1.616v. CPUID was most likely wrong, so your situation is normal.
Well, to not be all bitchy it seems that, indeed, all of these motherboards start with ~1.6v instead of 1.5v. On the other hand, I don't see how that's normal when there's no info about that anywhere. Not to mention that all budget DDR3 works at 1.5v if not overclocked. There's also the issue of being able to set 1.5v and it not being the real voltage.
Well, I'm not going to bother them further as I don't really expect them fixing this or providing me with more significant info. At least not in the next few months, probably ::).
Maybe I'll ask runn3R about this ;D
-
Well, I'm surprised by that response. On the GA-770TA-UD3 and on my own GA-790XTA-UD4 the default voltage is 1.5V for the RAM and it shows this correctly in both PC Health Status and CPUID.
It might be worth trying to get a proper answer from runn3R.
-
When I look at the PC Health Status page the vdimm is 1.616v. Everest is saying that the vdimm is 1.62v. I am using the latest F3 bios and all values are on auto. The DRAM voltage control in the BIOS starts only from 1.600v, but the manual says it should start from 1.500v
Same to me with any BIOS
1,62 V default value, 1666MHz, 8-8-8-20,2x2GB kingstone
But sometimes i see 1,5V in BIOS only; after booting everest report 1,62.I think it's a bug