Overclock.net › Forums › AMD › AMD Motherboards › [Official] Gigabyte GA-990FXA-Series Owners Thread/Club
New Posts  All Forums:Forum Nav:

[Official] Gigabyte GA-990FXA-Series Owners Thread/Club - Page 553

post #5521 of 12377
Quote:
Originally Posted by Baskt_Case View Post

Well, this board has presented it's own unique set of problems to me. At first yea, things were great, but I built this rig mainly for folding and it started crashing while Folding. It was previously stable while Folding for days on end, then just started crapping out at random.

So I'm working on that right now. I've started my OC from scratch and working my way through getting IBT and OCCT stable, then I will try folding again.

Luckily, through all of this, I have not experienced throttling, just general instability.

So far I am IBT Max stable, and 7 hours of OCCT at 4.6GHz @ 1.45v CPU @ 1.22v CPU NB @ 2.505v CPU PLL

Plenty of time to mess with all this, I'm off work on medical leave after I took a horrible laceration to my hand and nearly lost a finger. Couple broken bones too!

The print-out I'm working from says the CPU PLL is not really a factor and reccomends leaving it alone. which I did.

for the record, since the BIOS update still not a single POST-CHOKE, and in the last OC tweaks, no instability or throttling, course I only left it at 65 for about a minute and there has been NO load-testing yet.

This is what, (the guide) I'm (mainly) using:
http://www.overclock.net/t/1140459/bulldozer-overclocking-guide-performance-scaling-charts-max-ocs-ln2-results-coming
I originally found it at Tweaktown forums but I just now discovered we had it here first. (?)

Props to: Sin0822
post #5522 of 12377
Quote:
Originally Posted by Baskt_Case View Post

Check out this post I found on another forum. His Gigabyte board has 10 levels of LLC, and he did a test with a fixed BIOS vCore setting, and then booted using each LLC setting and logged the idle and load results.

One of of us needs to do this on our boards to get a definitive answer.

I've been using Auto since day one.

I guessing that's either a UD7, possibly a 5 and-or possibly an INTEL CS?
hmmmm. I know! maybe I should look at the link! thinking.gif
done.
yeah Ok it's an Intel. I have Z68 myself (not UD7) mines a 3 and I've had NO LUCK getting it OC AT ALL. It's pending a BIOS UPDATE. Still on out-the-box. hopefully that helps. Sure as heck did with the 990FXA.
Edited by NoNewEgg4Me - 5/18/13 at 12:38pm
post #5523 of 12377
I'll tell ya what, the only reason I bumped CPU PLL was because I failed OCCT after 1 hour. With a CPU PLL bump it ran for 7 hours. Thats all I know.

Thanks for linking that guide! I have not seen that one yet. Gonna grab a beer and give that a read.

And yea, that guys board is Intel. I just posted that for insight. Like I said, we need to test our boards like he did.
Edited by Baskt_Case - 5/18/13 at 12:37pm
    
CPUMotherboardGraphicsRAM
FX-8350 @ 4.6GHz 1.48v 990FXA-UD3 Pegatron HD6450 1GB 2x4GB G.Skill Sniper 1866 
Hard DriveOptical DriveCoolingOS
WD Blue 500GB Samsung DVD-RW Corsair H100i Win 7 x64 
MonitorPowerCase
LG 32" 720p HDTV SeaSonic M12II 620 Bronze CM 690 II Advanced 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
FX-8350 @ 4.6GHz 1.48v 990FXA-UD3 Pegatron HD6450 1GB 2x4GB G.Skill Sniper 1866 
Hard DriveOptical DriveCoolingOS
WD Blue 500GB Samsung DVD-RW Corsair H100i Win 7 x64 
MonitorPowerCase
LG 32" 720p HDTV SeaSonic M12II 620 Bronze CM 690 II Advanced 
  hide details  
Reply
post #5524 of 12377
does the ud5 990fx include dialblo iii
post #5525 of 12377
Quote:
Originally Posted by NoNewEgg4Me View Post

Very Interesting Update, ...for 6-core (FX6000 series) owners....

Since I'm one of a short list of 6core owners on this thread I thought you might find my recent results of interest, (by way of contrast) not to mention of actual USE to any 6core owners dropping by.

While I'm still waiting for my Coolmaster CPU cooler to arrive I decided it might be good to push a little further on NOW and 1.) get some idea how far I'm going to be able to go, and 2.) get a little preview of the temps when I do, partly in hopes of NOT having to pull the board out again which installing the in-the-mail cpu cooler requires.
Last I reported I was at X17.50 = 3.5Ghz on the CPU and 1866 on the ram. since then I'd already punched up the RAM to 1959, all good. So here's what I did today:
in order:
Voltages:
V-CORE: +.025
CPU NB VID +.025
NB: +0.02 (net 1.12v)
LLC, either "med" or I left it on AUTO, not sure (working from notes I'm not at home)
Freq's:
CPU HOST CLOCK to 210 MHZ
CPU FREQ up one tick to X18:00 = 3.6Ghz
REBOOT. POST OK, CPU was showing me only 3.53 (rougly) only a tiny lift.
Then I noticed the Vcore resultant voltage was only 0.958 and I knew I wanted more like, btwn 1.2-1.3.
So back to v-core voltage, made it +0.250 (up from the +.025) which by my math should have only made it 1.21v net.
REBOOT. Post-still-good!
WHAM: 3.78Ghz!! (mutilplier set to 3.6) w'hoo! but wait, what's the temp?
V-core temp showng, 65c just idling in BIOS! My underanding is I wanna stay under 60 if at all possible!
and V-CORE net voltage now fluct'ng btwn 1.54 and 1.6 also well over my target 1.3.
I hauled donkey quickly back to the VCORE voltage and cut the tweak by half to +0.125.
REBOOT, good post.
At first I was confused, CPU speed still showed as 3.78Ghz, but the V-CORE-TEMP was down to mid 40's! Sweet!
AND the V-CORE net voltage was now btwn-1.44 and 1.53 still a bit high but no longer scary-high.

Questions:
1.) I assume, based on these results the next move is to pull the V-CORE voltage back one tick at a time untill I see the 3.78Ghz speed flinch. The logic being you don't want to give the V-CORE any more power than whatever's just enough to support the 3.78. and see how low the temp drops in the process. correct?
2.) I assume also that a vcore temp of mid 40s AT IDLE is not Sweet-Home-Alabama and under any actual load could shoot up and past 60c with-in, well, don't blink, correct? soooo if I can get the vcore temp lower and still maitain the 3.78Ghz thats a good plan. Correct?
3.) Currently the multiplier is set to a limit of 3.6 but the current resultant speed is 3.78, assuming I wanted a nice round 3.8, is the following the best attack:
pull the Vcore voltage back untill the speed dips, then put it back to whatever it was last supported the 3.78, and kick the multiplier/freq up to 3.8

Here's an interesting side note, I re-assessed the Windows Experience after I got the temp down, but still at 3,78 and 1933/(59) on the RAM and the VIDEO SCORES had actually DROPPED from 7.1 to 7.0 (yet the CPU score was unchanged) It seemed like the new OC was now crowding the VIDEO for resources. True?

Soon I'm going to fill-in this and all my other builds. I know you kinda need to know PS's and all that other stuff to answer some of my questions.
Thanks Guys!

At that low of a clock speed I honestly think you shouldn't be anywhere near 1.5v on the vcore. I would by trying lower 1.4v's honestly.

A lower vcore will yield lower temps but also remember that higher clocks yield higher temps so they can compound each other. 40C idle in the BIOS is not going to be what it is in Windows, especially if you have C&Q enabled. However under a load it is possible that it will go above 60C just because it is that high in the BIOS.

The frequency being 3.78 sounds to me that either you have the multi 1 tick higher than you think or you have a slight FSB overclock going as well. What is CPUZ showing?
Quote:
Originally Posted by Baskt_Case View Post

I'll tell ya what, the only reason I bumped CPU PLL was because I failed OCCT after 1 hour. With a CPU PLL bump it ran for 7 hours. Thats all I know.

Thanks for linking that guide! I have not seen that one yet. Gonna grab a beer and give that a read.

And yea, that guys board is Intel. I just posted that for insight. Like I said, we need to test our boards like he did.

I have tried, Regular, Medium, and High. I have not noticed any difference in them honestly.
Quote:
Originally Posted by PCBuilder94 View Post

does the ud5 990fx include dialblo iii

I don't believe so...
The Postoffice
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700x Asus Prime X370-Pro EVGA 1080Ti Founders Edition G. Skill Ripjaws V  
Hard DriveHard DriveHard DriveOptical Drive
Kingston HyperX 3k WD Black 640 Samsung 850 EVO Sony Optiarc 
CoolingOSMonitorKeyboard
Corsair H100i Windows 10 64 Vizio VOJ320F1A Corsair K70 Vengance 
PowerCaseMouseAudio
NZXT Hale82 Antec TwelveHundred Logictech G9x Realtek 
  hide details  
Reply
The Postoffice
(16 items)
 
  
CPUMotherboardGraphicsRAM
AMD Ryzen 7 1700x Asus Prime X370-Pro EVGA 1080Ti Founders Edition G. Skill Ripjaws V  
Hard DriveHard DriveHard DriveOptical Drive
Kingston HyperX 3k WD Black 640 Samsung 850 EVO Sony Optiarc 
CoolingOSMonitorKeyboard
Corsair H100i Windows 10 64 Vizio VOJ320F1A Corsair K70 Vengance 
PowerCaseMouseAudio
NZXT Hale82 Antec TwelveHundred Logictech G9x Realtek 
  hide details  
Reply
post #5526 of 12377
Quote:
Originally Posted by Bkpizza View Post

Ok thanks heaps, gave it a crack, watched clocks with CPU-Z since I already had it and it stayed locked on 4633mhz the whole time. I'm pretty sure its ok because it was still doing 90ish GFLOPS in IBT...

Hey, I noticed that in an earlier post of yours as well. I'm testing 4.7GHz on an 8350 right now and the best I can muster is 45 GFLOPS. Whats up with that.

Is your version of IBT using AVX? I notice that your IBT looks a little different than mine. Although the version number is the same, 2.54.
    
CPUMotherboardGraphicsRAM
FX-8350 @ 4.6GHz 1.48v 990FXA-UD3 Pegatron HD6450 1GB 2x4GB G.Skill Sniper 1866 
Hard DriveOptical DriveCoolingOS
WD Blue 500GB Samsung DVD-RW Corsair H100i Win 7 x64 
MonitorPowerCase
LG 32" 720p HDTV SeaSonic M12II 620 Bronze CM 690 II Advanced 
  hide details  
Reply
    
CPUMotherboardGraphicsRAM
FX-8350 @ 4.6GHz 1.48v 990FXA-UD3 Pegatron HD6450 1GB 2x4GB G.Skill Sniper 1866 
Hard DriveOptical DriveCoolingOS
WD Blue 500GB Samsung DVD-RW Corsair H100i Win 7 x64 
MonitorPowerCase
LG 32" 720p HDTV SeaSonic M12II 620 Bronze CM 690 II Advanced 
  hide details  
Reply
post #5527 of 12377
Quote:
Originally Posted by Baskt_Case View Post

I'll tell ya what, the only reason I bumped CPU PLL was because I failed OCCT after 1 hour. With a CPU PLL bump it ran for 7 hours. Thats all I know.

Thanks for linking that guide! I have not seen that one yet. Gonna grab a beer and give that a read.

And yea, that guys board is Intel. I just posted that for insight. Like I said, we need to test our boards like he did.

no I absolutely agree we need someone to do that.
it was a good read.
post #5528 of 12377
Quote:
Originally Posted by OzzyRuleZ View Post

At that low of a clock speed I honestly think you shouldn't be anywhere near 1.5v on the vcore. I would by trying lower 1.4v's honestly.

A lower vcore will yield lower temps but also remember that higher clocks yield higher temps so they can compound each other. 40C idle in the BIOS is not going to be what it is in Windows, especially if you have C&Q enabled. However under a load it is possible that it will go above 60C just because it is that high in the BIOS.

The frequency being 3.78 sounds to me that either you have the multi 1 tick higher than you think or you have a slight FSB overclock going as well. What is CPUZ showing?
I have tried, Regular, Medium, and High. I have not noticed any difference in them honestly.
I don't believe so...
Thanks Oz,
Pretty much what I thought, I'm going to pull the VCORE Voltage back until I see the 3.78 dip or a post-fail.
AND I'm confident you're right about the 40c idle in bios not being golden. I thought so too. It's still good to hear someone else say it.

AS FOR The FSB OC? I only tweaked what I mentioned. Before that I'd only done XMP profile 1, memory manual multiplier, and the CPU freq-multiplier, (which I'm sure is at 18.00)
Like I said I was at 3.53 and the ONLY THING I changed was Vcore voltage from +.025 to +.250 and BAM~! 3.78, which, IMO is a pretty decent (high) clock speed for a PROC with stock speed of only 3.3, I'll be pleased as all get out, (this talking like we're all FIVE is hard!), to end up with a nice stable 3.8, (given the inherent design weaknesses of this board)

thank you and good nite!
Edited by NoNewEgg4Me - 5/18/13 at 2:39pm
post #5529 of 12377
This is for OZZ and everybody else having VRM overheating/warped Mobo's problems.
I could be wrong, though I don't believe I am. All you have to do...go to newegg and look at the pictures of the UD3, UD5 mobo's and look for the photo that shows the back of the board.
Look specifically on the board at the location of where the backside of the VRM's would be.....you will see nothing there, which is the location where the board warps from heat.

Now, at the newegg, look at a sabertooth board backside where the VRM's would be located and you will see a "support" that prevents the board from warping and the VRM's from losing contact with their heatsink.

That gentlemen...I believe is most of the problem. I sincerely hope this helps. My new Sabertooth will arrive on Tuesday. I wish you all well.
post #5530 of 12377
Quote:
Originally Posted by Pudfark View Post

This is for OZZ and everybody else having VRM overheating/warped Mobo's problems.
I could be wrong, though I don't believe I am. All you have to do...go to newegg and look at the pictures of the UD3, UD5 mobo's and look for the photo that shows the back of the board.
Look specifically on the board at the location of where the backside of the VRM's would be.....you will see nothing there, which is the location where the board warps from heat.

Now, at the newegg, look at a sabertooth board backside where the VRM's would be located and you will see a "support" that prevents the board from warping and the VRM's from losing contact with their heatsink.

That gentlemen...I believe is most of the problem. I sincerely hope this helps. My new Sabertooth will arrive on Tuesday. I wish you all well.


Well we knew the UD3 has warping issues and from what others have done "fixing" the warping and making the heat sink connect but since the heat sink sucks so much it makes no difference. This was just a dumb decision by Gigabyte, they tried to save a dollar and screwed the whole thing over in return.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD Motherboards
Overclock.net › Forums › AMD › AMD Motherboards › [Official] Gigabyte GA-990FXA-Series Owners Thread/Club