Overclock.net › Forums › Components › Memory › RAM - Same Brands, Different Speed/Timings, Help Please!
New Posts  All Forums:Forum Nav:

RAM - Same Brands, Different Speed/Timings, Help Please!

post #1 of 8
Thread Starter 
Hey everyone,

Been reading for a while. thumb.gif You've all helped me get the most out of all of my builds just from reading about other people's trials and such, but it's been difficult for me to find a clear answer on this question of mine--opinions I've read varied, so it's time that I just ask directly with my situation. Skim it if you have to, it has some unnecessary stuff I'm sure-- I'm trying to put in as much as possible that I deem remotely relevant...

Below are my specs:

CPU: AMD A10-7850k
M/B: Gigabyte GA-F2A88XM-D3H
RAM: See Below
PSU: Corsair AX860i 860W 80+ Platinum Fully-Modular ATX
Cooler: NZXT Kraken X60
GPU: Part of APU, unless the R7 250X works in Dual Graphics (I know the R7 250 does... wouldn't mind the Sapphire Vapor-X R7 250X OC for this at under $120)

Here's the RAM concern I have. Since I know the APU's GPU gets dramatic benefit from higher frequency RAM apparently even at the cost of timings, I have a little predicament.

I have 2 sets of 8GB DIMMs. One is newly purchased, the other "new" relative to most standards--they were already in my GA-F2A88XM-D3H before I acquired the 7850k (got sticks right before Christmas)

Set 1: Patriot Viper 3 Black Mamba - PV38G186C0K - 8GB (2x4GB) of DDR3-1866 - Box: 1.5V, 10-11-10-30.

This is what I had in my system already, when it was running with an A10-6800k. I never bothered tweaking this particular pair because they were just for my APU-based HTPC; with the new APU, I wanna change that. Although Gigabyte's website says 4 x 1.5v DDR3 DIMMs are supported, they're QVL includes the discontinued Patriot Viper Xtreme, Division 2, DDR3-2400 MHz at 11-11-11-30 @ 1.65v (PXD38G2400C11K), as well as the 2133 version at 11-11-11-27 @ 1.5v (PXD38G2133C11K). Also shows the Viper 3 Black Mamba DDR3-2133 16GB pair (just throwing it in for reference) on the QVL at 11-11-11-30 @ 1.5v.

My friend has essentially the same motherboard (he has the GA-F2A88XM-WIFI, instead) and the same exact RAM--we had purchased them together (cute, I know...). He was able to bump it up to 2133 MHz at CL11 though I don't have the exact voltage and timings on-hand. But it is stable.


Set 2: Patriot Viper 3 Black Mamba - PV38G240C0K - 8GB (2x4GB) of DDR3-2400 - Listed as: 1.65V, 10-12-12-30.

They're not only the same brand and tier within the brand, but also it would seem that they're essentially the same module types based on the product number. The 2400 has had some good reviews for overclocking, with pureoverclock's review apparently hitting 2800 MHz at 11-14-14-36 1T, and even getting it to boot into windows at 2884 MHz at 11-14-14-36 1T but not stable. The AnandTech Review has it at 2666 MHz @ 10-12-12-?? @ 1.72v.

Either way...I couldn't find much on the 1866 pair, which is the product I'm more focused on for this question. While my friend (as noted above) has seen some great results, you really never know until you try it with your pair, and I want to know if I should even bother trying this little endeavor...

So, finally, here is my question... my concern stems from the fact that the listed Base Timings confuse me a bit. The 1866 MHz says 1600 MHz @ 9-9-9-24 @ 1.5v, while the 2400 has 1600 @ 11-11-11-28 @ 1.65v.

I want to be able to run both pairs, filling up all 4 DIMMs, at 2400 MHz if possible, and 2133 if not. The motherboard supports the speed. Otherwise, I'd probably end up buying another pair of the DDR3-2400 set...but I'd rather save my money if possible, considering I would be purchasing more RAM and making 2 other sticks essentially worthless. If the "individual pairs" (Set 1 vs Set 2) can be stable at the same voltage and timings, will this function properly given the above information?

I'm not concerned about my ability to get to a stable overclock on Set 1, as much as compatibility with Set 2.

While I am not an idiot at tweaking computers and understand the importance of voltage, latency, etc., I'm no where nearly as experienced as some of you, and won't claim to be just by writing numbers on a post! I've never tried to do this particular type of tweaking with a DDR3 set, and just want to see if it at least makes sense...especially considering this is an APU I haven't had much experience working with, like most (I assume...) people currently.

Kaveri aside, if someone can explain whether or not it is even possible to achieve my goal, please let me know! I have looked at the Kaveri forum thread extensively, so I would greatly appreciate not simply being told to post it there or read about it there. Since this is a memory-specific question, and I don't want to post an essay on an on-going thread, I decided to start it here. All other links to anything that could educate me, however, I am more than open to (both about Kaveri and RAM OC'ing).

Thanks for reading all this, hopefully it's less annoying than those who just say "i have x ram how can I overclock to make my pc awesome" biggrin.gif I've seen a few of those...

- Mamba (ironically, this has been my nickname...and happens to be the RAM I have...go figure...)
post #2 of 8
Quote:
Originally Posted by MrMamba View Post

So, finally, here is my question... my concern stems from the fact that the listed Base Timings confuse me a bit. The 1866 MHz says 1600 MHz @ 9-9-9-24 @ 1.5v, while the 2400 has 1600 @ 11-11-11-28 @ 1.65v.

I want to be able to run both pairs, filling up all 4 DIMMs, at 2400 MHz if possible, and 2133 if not.

Perhaps this will help:

Memory timings are the amount of time (in nano-seconds) that it takes the memory to complete that step in what it has to do (read,write, or move data).
The number (ie the nunber 9 in cas 9, for example) is the number of computer clock ticks it takes to finish that step (ie, at 4 GHZ = 0.00000000025 seconds per clock tick, 1/4000000000). If you run the memory at a faster speed it still takes the same amount of time to complete the cas step ( 9 x 0.00000000025 seconds). But, since each clock tick is shorter, it takes more of then to complete the step. Going from 4 GHZ to 4.5 GHZ for example, 4.5 GHZ/4.0 GHZ x 9 = 10.25 clock ticks (rounded to 11 as it has to be a whole number of clock ticks).

So, basically, at 1600 mhz cas11 takes a much longer time to complete the cas step than does 1600 cas 9.
You can run the 1600 cas 9 at cas 11, but you can't run the 1600 cas 11 at cas 9. There just isn't enough clock ticks at cas 9 as 9 clock ticks just isn't enough time to complete that step.

When mixing memory, the calculations get tedious as there are about 28 different timings that effect performance and stability (even though the manufacturers only list 5 or 6 of them).

Hope that helps.
Edited by billbartuska - 2/18/14 at 6:25am
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
post #3 of 8
Thread Starter 
Quote:
Originally Posted by billbartuska View Post

Perhaps this will help:
...

So, basically, at 1600 mhz cas11 takes a much longer time to complete the cas step than does 1600 cas 9.
You can run the 1600 cas 9 at cas 11, but you can't run the 1600 cas 11 at cas 9. There just isn't enough clock ticks at cas 9 as 9 clock ticks just isn't enough time to complete that step.

When mixing memory, the calculations get tedious as there are about 28 different timings that effect performance and stability (even though the manufacturers only list 5 or 6 of them).

Hope that helps.

It does, but what I'm concerned about is whether or not I can use both sticks simultaneously if, by chance, I can get the 1866 set to operate at 2133 or 2400 with the same timings as the 2400 pair. PureOC's review shows that they were able to get it down to 7-8-7-24, so if the base timings were 1600 MHz at 11-11-11-28, they were clearly able to get a CAS 11 to CAS 7, so why couldn't they run it at CAS 9? Forgive my ignorance here.

I just want to know that if these two sticks can be adjusted enough to work together at either 2133 or 2400. Simultaneously... hope that narrows down the argument.
post #4 of 8
Quote:
Originally Posted by MrMamba View Post

I just want to know that if these two sticks can be adjusted enough to work together at either 2133 or 2400. Simultaneously... hope that narrows down the argument.

Well, it does.
The answer is maybe, but not likely.
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
post #5 of 8
Thread Starter 
Quote:
Originally Posted by billbartuska View Post

Well, it does.
The answer is maybe, but not likely.

"So you're saying there's a chance..." haha... well, okay. Is this based on the unlikelihood of getting exactly identical stable timings for both, or some other variable I haven't considered?

Thank you, by the way.
post #6 of 8
My guess would be:
Run together at 1600 - 50% chance
Run together at 2400 - 0% chance.

A lot will depend on how many timings your BIOS will allow you to change.

Edit: Checked your owner's manual. You can change 13 of the 28 timings...not good.
Edited by billbartuska - 2/18/14 at 11:49pm
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
My System
(15 items)
 
  
CPUMotherboardGraphicsRAM
FX6300 Black M5A99X EVO R2.0 Nvidia GTS450 Team Vulcan PC3 12800 
Hard DriveOptical DriveCoolingOS
Samsung 840 PRO Asus DRW-1608P (x2) Custom Water Cooling Win7 (Ult), Win 8.1 & Win Server 2012 R2 
MonitorKeyboardPowerCase
2 X Samsung 915N Ducky Shine III, Blue Cherry/Blue LEDs PCP&C 1kw Lian Li PC-71 (W/Window) 
MouseAudio
Logiteck G400s none 
  hide details  
Reply
post #7 of 8
Thread Starter 
Quote:
Originally Posted by billbartuska View Post

My guess would be:
Run together at 1600 - 50% chance
Run together at 2400 - 0% chance.

A lot will depend on how many timings your BIOS will allow you to change.

Edit: Checked your owner's manual. You can change 13 of the 28 timings...not good.

Fun, guess I'll find another use for the 1866 sticks and get another pair of the 2400's. Thank you for your help.
post #8 of 8
Thread Starter 
Quote:
Originally Posted by billbartuska View Post

My guess would be:
Run together at 1600 - 50% chance
Run together at 2400 - 0% chance.

A lot will depend on how many timings your BIOS will allow you to change.

Edit: Checked your owner's manual. You can change 13 of the 28 timings...not good.

So I went ahead and got a 2nd set of the 2400, they actually just exchanged them since the sticks were the same price...not even open box exchange... quite pleasantly surprised. I ran with 8GB at 2400 for the last week or so, no issues whatsoever, which makes me happy. (I copied the XMP profile settings, then manually re-input them so I can tweak as needed, which brings me to the points below--plus I still am not completely confident in the XMP being fully supported by AMD mobos...I'm likely wrong...)

Thanks for all the help, one final question!

I just installed the second set of modules. Channel A shows me timings of 10-12-12-30, while channel B shows 10-12-12-31 (although the Patriot box indicates 10-12-12-30) with the XMP profiles. CPU-Z shows me 10-13-13-32, if that is of any consequence.

The part that confuses me a bit more is that in the advanced timings.

Channel A tRC: 43
But I was under the impression that tRC was tRAS + tRP.
Channel A tRAS + tRP = 42, yet it shows 43 for tRC, but has been running fine. No hangup with hours and hours of normal use and heavy loads, all spread across days without reboot.

Channel B tRC: 43
Channel B tRAS + tRP = 43, which is consistent with tRC of 43.
Unless I'm completely mistaken on this formula, I'm pretty sure my mathematics add up... or the BIOS is rounding fractions...ha...

Channel A tRFC0: 4; tRFC1: 4
Channel B tRFC0: 3; tRFC1: 3
I can't put my finger on why the Refresh to Act Delay would be different on two identical modules... I know I'm missing something.

The Address/Command Timing is different slightly, as well--
Channel A CKE fine delay: 12
Channel B CKE fine delay: 20

They are both the same 2x4GB set--both boxes indicate 10-12-12-30 (as do the specifications on Patriot's site). Any thoughts?

Again, apologies for sounding like a naive person... I really do know a little about this, but my knowledge with systems is in different areas and I've never really needed to expand it into this--but I am now. =) I plan to get some CPU and GPU OC'ing underway once I get this going. Thanks again!
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Memory
Overclock.net › Forums › Components › Memory › RAM - Same Brands, Different Speed/Timings, Help Please!