Overclock.net › Forums › AMD › AMD CPUs › Overclocking the FX-8320E with a Gigabyte 990FXA-UD5 Rev3.0
New Posts  All Forums:Forum Nav:

Overclocking the FX-8320E with a Gigabyte 990FXA-UD5 Rev3.0 - Page 2

post #11 of 154
Thread Starter 
Quote:
Originally Posted by BenC View Post

Keep in mind the lower TDP is a product of the lower core speed/ vCore. Once you bump the vCore up the TDP goes out the window.

I need to do some systematic testing (better yet find several people do systematic testing) but I've noticed a few things with my 8370E. First, the chip and my NB run hotter than my 8350 did with lower vCore once I go past stock. My memory is running considerably faster (~1800 --> 2133) so that may explain the NB. At 4.5 @1.42 the 8350 would peak around 50C, where as 4.7 @ 1.39 on the 8370E goes to ~ 54 or higher. Obviously delta is more important but I don't have those numbers.

Second, undervolting seems to be about the same as other 8350s (mine was a bit subpar), with 4.5 around 1.32. 4 @ 1.17 does seem to be an improvement in comparison (again, not a systematic comparison). I do see an improvement with overclocking. For 4.7 my 8350 required 1.46 and could not handle memory above 1800. Any increases beyond 4.7 took big voltage increases. So far I've topped out at 1.39 for 4.7; basically stock voltage of my old 8350.

Finally, the IMC is improved, maybe not to every 8350, but compared to the 8350 I had.

Unfortunately it is difficult to do a large comparison because of different hardware, etc. And that I just plan do not believe most of the numbers in the 8350/8320 club.

Thank you for this. The #'s in the 8350/8320 club are all over the place in that spreadsheet. I can't even begin to use anything in there. I will likely get some more time later this week (weekend too) to figure out the voltage/thermal sweet spots for the FX-8320E... Until then, I will just be taking some wild swings to see if I get lucky. Anything above 1.4V on these E chips -- things get really hot, fast.. I saw that too.

What LLC levels were you rolling with? With Ultra High LLC, I am seeing more vDroop than I did with the FX-4130. It's not terrible, but I am seeing some difference.

Based on my reading, I am going to ditch Prime95 in lieu of the IBT to do faster stability testing. I don't mind doing a 12 hour burn when I go to bed.. but when I'm actively trying to find stable points, 10 minutes is better than 3-12 hours.
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
post #12 of 154
It could be you're at the dissipation limit for the 212.
Lil' Roy Taylor
(11 items)
 
  
Reply
Lil' Roy Taylor
(11 items)
 
  
Reply
post #13 of 154
Quote:
Originally Posted by ultrasparc View Post

Anything above 1.4V on these E chips -- things get really hot, fast.. I saw that too.

Lapping and multiple reseats of the cooler helped. I took mine off when I switched back to water but a fan blowing on the backside of the socket helps quite a bit. I know there is a thread or multiple posts about this somewhere.
Quote:
What LLC levels were you rolling with? With Ultra High LLC, I am seeing more vDroop than I did with the FX-4130. It's not terrible, but I am seeing some difference.

I believe I have been using high LLC across the board. I'll have to double check tonight when I get home. I am certain my vCore is a little higher outside of stressing than it would be, but closer to where I intend it to be during stressing than Ultra or Extreme would put it.
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
post #14 of 154
Thread Starter 
Quote:
Originally Posted by p4inkill3r View Post

It could be you're at the dissipation limit for the 212.

That is quite possible.
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
post #15 of 154
Thread Starter 
Quote:
Originally Posted by BenC View Post

Lapping and multiple reseats of the cooler helped. I took mine off when I switched back to water but a fan blowing on the backside of the socket helps quite a bit. I know there is a thread or multiple posts about this somewhere.
I believe I have been using high LLC across the board. I'll have to double check tonight when I get home. I am certain my vCore is a little higher outside of stressing than it would be, but closer to where I intend it to be during stressing than Ultra or Extreme would put it.

I will have to try re-seating it at some point.

About the LLC -- I will try High when I get home and pay attention to where the voltage lands as the testing occurs.. Obviously we don't care about a lower idle voltage, but it must be relatively stable with load at the right voltage. Ultra High is putting me just a smidge too high I think.
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
post #16 of 154
Quote:
Originally Posted by BenC View Post


Unfortunately it is difficult to do a large comparison because of different hardware, etc. And that I just plan do not believe most of the numbers in the 8350/8320 club.

My idea of the different numbers in the Vishera club, is that... they are true. There is an extreme variability between chips. Some of the early ones must have been particularly high leakers, draining humongoys amounts of watts from the wall (there is an anandtech member that on stock was having 200W+ draw only from the CPU). My FX6300 is as bad as it gets. Needs 1.334v for 4Ghz and overclocks like a dog. My 8320 does 1.256v for 4Ghz and undervolted draws exactly the same as my power hungry 1090T (undervolted). So, AMD is throwing in a bit all kind of chips.

AMD can't afford to trash chips only because they are marginal or even plain power hungry, exceeding nominal TDP somewhat. The luck of the draw is true for the Visheras more than before. *Some* results in the Vishera Club might be completely off, due to the fact that some people don't really stress test seriously their CPU, but are happy say with 30 minutes Prime. Obviously when you mix such results, with others that run Prime for 10 hours, you end up with some misleading results. But the variability between chips, exists and it's significant.
Main
(16 items)
 
Dedicated Encoder
(15 items)
 
 
CPUMotherboardGraphicsRAM
FX-8320@4Ghz Gigabyte 970 UD3P rev2.1 Gainward GTX 750Ti Corsair XMS3 1600Mhz 16GB (4x4GB) 
Hard DriveHard DriveOptical DriveOptical Drive
Crucial BX100 250GB Western Digital Green 2TB LiteOn Blu-Ray Burner IHBS 112-2 LG BH16NS55 Blu-Ray Burner 
CoolingOSMonitorKeyboard
Scythe Katana 3 Windows 7 Pro 64bit ASUS 22" VS228HR Microsoft Wired Keyboard 600 
PowerCaseMouseAudio
EVGA 430W Sharkoon VG4-V Logitech M90 Onboard 
CPUMotherboardGraphicsRAM
FX-8300 Asrock 970 Extreme3 HIS 6570 Silence Corsair XMS3 1600Mhz 8GB (2x4GB) CAS9 
Hard DriveHard DriveOptical DriveCooling
Plextor M6S 128GB Toshiba 2TB SATAIII LiteOn Blu Ray burner IHBS 112-2 Xigmatek Balder 
OSMonitorKeyboardPower
Windows 7 Pro 64bit Samsung S22B350H Microsoft Wired 600 Corsair VS350 
CaseMouseAudio
Lepa LPC 306 Logitech M90 Onboard 
  hide details  
Reply
Main
(16 items)
 
Dedicated Encoder
(15 items)
 
 
CPUMotherboardGraphicsRAM
FX-8320@4Ghz Gigabyte 970 UD3P rev2.1 Gainward GTX 750Ti Corsair XMS3 1600Mhz 16GB (4x4GB) 
Hard DriveHard DriveOptical DriveOptical Drive
Crucial BX100 250GB Western Digital Green 2TB LiteOn Blu-Ray Burner IHBS 112-2 LG BH16NS55 Blu-Ray Burner 
CoolingOSMonitorKeyboard
Scythe Katana 3 Windows 7 Pro 64bit ASUS 22" VS228HR Microsoft Wired Keyboard 600 
PowerCaseMouseAudio
EVGA 430W Sharkoon VG4-V Logitech M90 Onboard 
CPUMotherboardGraphicsRAM
FX-8300 Asrock 970 Extreme3 HIS 6570 Silence Corsair XMS3 1600Mhz 8GB (2x4GB) CAS9 
Hard DriveHard DriveOptical DriveCooling
Plextor M6S 128GB Toshiba 2TB SATAIII LiteOn Blu Ray burner IHBS 112-2 Xigmatek Balder 
OSMonitorKeyboardPower
Windows 7 Pro 64bit Samsung S22B350H Microsoft Wired 600 Corsair VS350 
CaseMouseAudio
Lepa LPC 306 Logitech M90 Onboard 
  hide details  
Reply
post #17 of 154
Quote:
About the LLC -- I will try High when I get home and pay attention to where the voltage lands as the testing occurs.. Obviously we don't care about a lower idle voltage, but it must be relatively stable with load at the right voltage. Ultra High is putting me just a smidge too high I think.

Exactly, C6 and Cool'n Quiet will take care of lowering idle voltage once you get a good OC.
Quote:
My idea of the different numbers in the Vishera club, is that... they are true. There is an extreme variability between chips. Some of the early ones must have been particularly high leakers, draining humongoys amounts of watts from the wall (there is an anandtech member that on stock was having 200W+ draw only from the CPU). My FX6300 is as bad as it gets. Needs 1.334v for 4Ghz and overclocks like a dog. My 8320 does 1.256v for 4Ghz and undervolted draws exactly the same as my power hungry 1090T (undervolted). So, AMD is throwing in a bit all kind of chips.

AMD can't afford to trash chips only because they are marginal or even plain power hungry, exceeding nominal TDP somewhat. The luck of the draw is true for the Visheras more than before. *Some* results in the Vishera Club might be completely off, due to the fact that some people don't really stress test seriously their CPU, but are happy say with 30 minutes Prime. Obviously when you mix such results, with others that run Prime for 10 hours, you end up with some misleading results. But the variability between chips, exists and it's significant.

You are right, the numbers certainly have a wide array of variables that influence them and variability between chips exist. But within the top 5 there is an entry for 4.6 @ 1.35! (Edit: Not the best example and is certainly possible.) I am not saying those numbers are useless but I think the outliers make it difficult to accurately gauge what a user could expect.

I'll make a note to group those numbers by similar hardware and calculate a few statistics (this may be a couple weeks).
Edited by BenC - 10/21/14 at 7:55am
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
post #18 of 154
Quote:
Originally Posted by BenC View Post


You are right, the numbers certainly have a wide array of variables that influence them and variability between chips exist. But within the top 5 there is an entry for 4.6 @ 1.35! (Edit: Not the best example and is certainly possible.) I am not saying those numbers are useless but I think the outliers make it difficult to accurately gauge what a user could expect.

I'll make a note to group those numbers by similar hardware and calculate a few statistics (this may be a couple weeks).

This is what happens when there is no standardization. For example, one may stress with IBT standard (much easier to pass), another with Prime 30 miniutes, another with Prime 1h, another with AIDA. One may write as voltage the setting he put in BIOS, another the effective (load) in Windows. For example, one may put 1.35v, meaning he put 1.35 in BIOS, but with LLC, it may have been higher in real load. There is no standardization, so you can't trust everything. I trust mainly the "stock" VID. Beyond that, once everyone stresses with whatever he likes, you can't rely on everything to be valid.
Main
(16 items)
 
Dedicated Encoder
(15 items)
 
 
CPUMotherboardGraphicsRAM
FX-8320@4Ghz Gigabyte 970 UD3P rev2.1 Gainward GTX 750Ti Corsair XMS3 1600Mhz 16GB (4x4GB) 
Hard DriveHard DriveOptical DriveOptical Drive
Crucial BX100 250GB Western Digital Green 2TB LiteOn Blu-Ray Burner IHBS 112-2 LG BH16NS55 Blu-Ray Burner 
CoolingOSMonitorKeyboard
Scythe Katana 3 Windows 7 Pro 64bit ASUS 22" VS228HR Microsoft Wired Keyboard 600 
PowerCaseMouseAudio
EVGA 430W Sharkoon VG4-V Logitech M90 Onboard 
CPUMotherboardGraphicsRAM
FX-8300 Asrock 970 Extreme3 HIS 6570 Silence Corsair XMS3 1600Mhz 8GB (2x4GB) CAS9 
Hard DriveHard DriveOptical DriveCooling
Plextor M6S 128GB Toshiba 2TB SATAIII LiteOn Blu Ray burner IHBS 112-2 Xigmatek Balder 
OSMonitorKeyboardPower
Windows 7 Pro 64bit Samsung S22B350H Microsoft Wired 600 Corsair VS350 
CaseMouseAudio
Lepa LPC 306 Logitech M90 Onboard 
  hide details  
Reply
Main
(16 items)
 
Dedicated Encoder
(15 items)
 
 
CPUMotherboardGraphicsRAM
FX-8320@4Ghz Gigabyte 970 UD3P rev2.1 Gainward GTX 750Ti Corsair XMS3 1600Mhz 16GB (4x4GB) 
Hard DriveHard DriveOptical DriveOptical Drive
Crucial BX100 250GB Western Digital Green 2TB LiteOn Blu-Ray Burner IHBS 112-2 LG BH16NS55 Blu-Ray Burner 
CoolingOSMonitorKeyboard
Scythe Katana 3 Windows 7 Pro 64bit ASUS 22" VS228HR Microsoft Wired Keyboard 600 
PowerCaseMouseAudio
EVGA 430W Sharkoon VG4-V Logitech M90 Onboard 
CPUMotherboardGraphicsRAM
FX-8300 Asrock 970 Extreme3 HIS 6570 Silence Corsair XMS3 1600Mhz 8GB (2x4GB) CAS9 
Hard DriveHard DriveOptical DriveCooling
Plextor M6S 128GB Toshiba 2TB SATAIII LiteOn Blu Ray burner IHBS 112-2 Xigmatek Balder 
OSMonitorKeyboardPower
Windows 7 Pro 64bit Samsung S22B350H Microsoft Wired 600 Corsair VS350 
CaseMouseAudio
Lepa LPC 306 Logitech M90 Onboard 
  hide details  
Reply
post #19 of 154
Thread Starter 
BenC -- Interesting, we are using the same case basically.. I bought 3 of the NZXT Source 210 Elite to get all my motherboards & CPUs in cases. I think it is a great case compared to some of the garbage I have bought in the past. I really need to go through the Rig Builder, lol.

Have you added a side fan to yours? I found there was definitely a slight thermal advantage when I added a 120mm to the side. I ordered a 140mm Corsair to place there instead. I couldn't fit a fan on the bottom.. My 1300W EVGA PSU is enormous. Of course you use water cooling, so I'm not sure it would matter as much to you, as I am air cooled... Might help with the NB though.
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
Primary - "Annie"
(12 items)
 
Backup - "Bena"
(15 items)
 
 
CPUMotherboardGraphicsHard Drive
AMD Athlon II X2 240 MSI 785GT-E63 ASUS R9270-DC2OC-2GD5 Radeon R9 270 2GB 256-bit Intel® SSD X25-E Series 30GB 
Hard DriveOptical DriveOSPower
WD Caviar Blue 640GB LG Electronics Optical DVD Drive GH24NSB0B Windows 7 x64 OCZ Technology ModXStream Pro 600W Modular ATX ... 
Case
NZXT Source 210 Elite 
  hide details  
Reply
post #20 of 154
Quote:
Originally Posted by Undervolter View Post

This is what happens when there is no standardization. For example, one may stress with IBT standard (much easier to pass), another with Prime 30 miniutes, another with Prime 1h, another with AIDA. One may write as voltage the setting he put in BIOS, another the effective (load) in Windows. For example, one may put 1.35v, meaning he put 1.35 in BIOS, but with LLC, it may have been higher in real load. There is no standardization, so you can't trust everything. I trust mainly the "stock" VID. Beyond that, once everyone stresses with whatever he likes, you can't rely on everything to be valid.

In this case the numbers are divided by idle and load but! It is a standardization issue and that is not what those numbers are for. So I am not saying those numbers are useless but if they are being used to gauge the potential of similar hardware may not be an accurate reflection.

It would be nice to create a template i.e. multi only, HT and NB at this, mem at this, split by: cooler, tim, brand, and bios. Then have a clock-off. That would reduce some of the variability, and I think, shed a little light on improvement to the manufacturing process and variability between chips.
Quote:
Have you added a side fan to yours? I found there was definitely a slight thermal advantage when I added a 120mm to the side. I ordered a 140mm Corsair to place there instead. I couldn't fit a fan on the bottom.. My 1300W EVGA PSU is enormous. Of course you use water cooling, so I'm not sure it would matter as much to you, as I am air cooled... Might help with the NB though.

It is a good low-cost case! I switched the side panels and mounted a 120 on the outside of the vented panel. If you invert it, the venting doesn't quite line up with the back of the socket but it is pretty close.

Edit: lack of coffee
Edited by BenC - 10/21/14 at 8:50am
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
Moneypit
(11 items)
 
  
CPUMotherboardGraphicsRAM
FX-8370E ASUS Sabertooth MSI 7950 G SKILL Sniper 
Hard DriveHard DriveHard DriveCooling
Agility 2 Agility 3 Western Digital Black  Custom Loop 
OSPowerCase
Win 7 Pro Corsair 750W NZXT 
  hide details  
Reply
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
Overclock.net › Forums › AMD › AMD CPUs › Overclocking the FX-8320E with a Gigabyte 990FXA-UD5 Rev3.0