Help with Ballistix Memory

Toggle sidebar Toggle sidebar
M

mtl77733

Member
Sep 12, 2013
33
0
66
I bought this Ballistix DDR3-1600 memory here - http://www.crucial.com/usa/en/blt2kit4g3d1608dt1tx0 and I have a question. It doesn't work at 800 MHz 8-8-8-24-36-2T timings for my mobo (Asrock M3A790GXH/128M) because my mobo requires overclocking to be able to use DDR3-1600. So, without overclocking, my next choice is 666 MHz = DDR3-1333. Can I use this memory with timings of 8-8-8-24-36-2T for 666 MHz? If not, then what timings do you suggest for 666 MHz?

Also I have another question. This Ballistix memory runs at 1.5V but I can't select 1.5V in my BIOS. The only options that are close to 1.5V are 1.48V and 1.53V. Can I use 1.53V? It's not too much voltage - only 0.03V higher, right? Do you think it's too much? I hope not.

Thanks in advance for your much appreciated help!
 
Last edited:
LTC8K6

LTC8K6

Lifer
Mar 10, 2004
28,520
1,575
126
I think it will easily run at 800/1600 if you set the memory clock to 800 in the BIOS and leave the other settings on auto.

The voltage is fine.
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
LTC8K6 said:
I think it will easily run at 800/1600 if you set the memory clock to 800 in the BIOS and leave the other settings on auto.

The voltage is fine.
Click to expand...

If I do that, I am unable to boot into Windows. My mobo is an old one that does not support DDR3-1600 unless the CPU is overclocked. I don't want to have to overclock the CPU to use my RAM at DDR3-1600.

Good to know that the voltage is fine.

Thanks!
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
I set the memory at 666 MHz with timings of 8-8-8-24-36-2T and tested it with MemTest86+. On the first 3 passes there were no errors. On the 4th pass there were two errors with only one bit not matching the correct data bits. On the 5th and 6th passes there were no errors. These tests took about 9 hours total.

Do you think this result is good enough to conclude that the memory is working fine at 8-8-8-24-36-2T for 666 MHz? I mean, only two errors in a very long test - can I safely assume that these errors are very unlikely to occur in real life operation of the computer?
 
William Gaatjes

William Gaatjes

Lifer
May 11, 2008
22,220
1,411
126
mtl77733 said:
I set the memory at 666 MHz with timings of 8-8-8-24-36-2T and tested it with MemTest86+. On the first 3 passes there were no errors. On the 4th pass there were two errors with only one bit not matching the correct data bits. On the 5th and 6th passes there were no errors. These tests took about 9 hours total.

Do you think this result is good enough to conclude that the memory is working fine at 8-8-8-24-36-2T for 666 MHz? I mean, only two errors in a very long test - can I safely assume that these errors are very unlikely to occur in real life operation of the computer?
Click to expand...

The mantra should be : An error is fail.
You must absolutely have zero errors.
See it like this, you load up an mp3 in memory with an editor , one bit gets corrupted without you knowing it and you write it back to the HDD or SSD.
And every time this happens. After while , suddenly your mp3 had pops and clicks in it.This can happen to every file you load in memory. Savegames could become corrupt. Games would randomly crash for no reason. WIndows blue screening.
If you can , start increasing the timings a bit. That will make it slower but it is a first step in 100% reliability.
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
William Gaatjes said:
The mantra should be : An error is fail.
You must absolutely have zero errors.
See it like this, you load up an mp3 in memory with an editor , one bit gets corrupted without you knowing it and you write it back to the HDD or SSD.
And every time this happens. After while , suddenly your mp3 had pops and clicks in it.This can happen to every file you load in memory. Savegames could become corrupt. Games would randomly crash for no reason. WIndows blue screening.
If you can , start increasing the timings a bit. That will make it slower but it is a first step in 100% reliability.
Click to expand...

We need to qualify how many passes for which we consider an error as fail. How many passes is enough? Because we can do 100 passes which will take a very long time and if an error occurs on the 100th pass then we consider it as fail. But on the other hand, if we do only 70 passes and no error happens and we're satisfied with that, then we consider it as success. So the 100 passes is fail but the 70 passes is success.

Come to think of it, we can never really be 100% certain of success unless we do a very high number of passes, say 10,000 passes, but do we have time for that? So, for practicality, we need to set a limit as to what is good enough or reliable enough, as nothing can really be 100% reliable in our imperfect world.

If I'm going to increase the timings a bit, which timing parameter should I start increasing first (CL or tRAS or tRCD or tRP or tRC, etc)? Which timing parameter is the most sensitive?
 
LTC8K6

LTC8K6

Lifer
Mar 10, 2004
28,520
1,575
126
mtl77733 said:
If I do that, I am unable to boot into Windows. My mobo is an old one that does not support DDR3-1600 unless the CPU is overclocked. I don't want to have to overclock the CPU to use my RAM at DDR3-1600.

Good to know that the voltage is fine.

Thanks!
Click to expand...
What CPU?

Seems odd that you can't run the ram at it's rated speed at least.

Every board lists the ram speeds above standard as an overclock, so that is not unusual.

Note that a new gigabyte Z370 board lists every mem speed above standard as an overclock.

https://www.gigabyte.com/Motherboard/Z370-AORUS-Gaming-5-rev-10#sp
 
Last edited:
M

mtl77733

Member
Sep 12, 2013
33
0
66
LTC8K6 said:
http://www.asrock.com/mb/memory/M3A790GXH128M.pdf

It will run at least some ram at 1600
Click to expand...

The CPU is an AMD Phenom II X3 720 BE.

That PDF you linked to is only telling half the story. It's just a list of tested compatible memory but it does not say if the DDR3-1600 modules in the list will run without overclocking the CPU.

Please see the specs here...

https://www.asrock.com/mb/AMD/M3A790GXH128M/index.asp

In the memory section it says, "Supports DDR3 1600(OC)/1333/1066/800 non-ECC, un-buffered memory". Note the "OC" for DDR3 1600, which means the CPU must be overclocked in order to be able to use DDR3-1600.

Also, another problem is that my BIOS actually doesn't have exact 800 MHz as an option. There is 667 MHz and after that is 833 MHz. 833 MHz is too much so I can't use it even if I overclock.
 
William Gaatjes

William Gaatjes

Lifer
May 11, 2008
22,220
1,411
126
mtl77733 said:
We need to qualify how many passes for which we consider an error as fail. How many passes is enough? Because we can do 100 passes which will take a very long time and if an error occurs on the 100th pass then we consider it as fail. But on the other hand, if we do only 70 passes and no error happens and we're satisfied with that, then we consider it as success. So the 100 passes is fail but the 70 passes is success.

Come to think of it, we can never really be 100% certain of success unless we do a very high number of passes, say 10,000 passes, but do we have time for that? So, for practicality, we need to set a limit as to what is good enough or reliable enough, as nothing can really be 100% reliable in our imperfect world.

If I'm going to increase the timings a bit, which timing parameter should I start increasing first (CL or tRAS or tRCD or tRP or tRC, etc)? Which timing parameter is the most sensitive?
Click to expand...

Well, in less than 9 hours you had multiple errors. That would qualify as fail.
I myself test for at least 18 hours minimum and absolutely no errors allowed.
If you can run it for 18 hours ( in one go or two passes of 9 hours) without errors, it seems less likely you will have issues. It is a start.
It is your own choice to allow errors, but when your pc suddenly freezes or blue screens while you are in the middle of something very important... It is your choice.
It is not as if you are living in space or living on top of a plutonium storage that you get these errors because of radiation interference.
I am no memory overclocking expert at all. What you can do is increase all to 9-9-9-26-38-2T. Just increase it a bit and test it.
I assume that when having very relaxed timings, it should be possible to test if it is a timings issue or a defect dram chip.
I once had a memory kit where one of the dimms was defective. Even with very relaxed timings, it would fail once in a while after a an hour. If that is the case, one of the memory chips on the dimm is just bad and you cannot use it.
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
William Gaatjes said:
Well, in less than 9 hours you had multiple errors. That would qualify as fail.
I myself test for at least 18 hours minimum and absolutely no errors allowed.
If you can run it for 18 hours ( in one go or two passes of 9 hours) without errors, it seems less likely you will have issues. It is a start.
It is your own choice to allow errors, but when your pc suddenly freezes or blue screens while you are in the middle of something very important... It is your choice.
It is not as if you are living in space or living on top of a plutonium storage that you get these errors because of radiation interference.
I am no memory overclocking expert at all. What you can do is increase all to 9-9-9-26-38-2T. Just increase it a bit and test it.
I assume that when having very relaxed timings, it should be possible to test if it is a timings issue or a defect dram chip.
I once had a memory kit where one of the dimms was defective. Even with very relaxed timings, it would fail once in a while after a an hour. If that is the case, one of the memory chips on the dimm is just bad and you cannot use it.
Click to expand...

Thanks, man!

Before I started this thread, I already tested 9-9-9-24-33-2T for 9 hours and it passed without any errors. I just want to see if I can improve the timings any further. So, if 8-8-8-24-36-2T is not working perfectly, I'm hoping someone could guide me as to what timings to try - maybe 8-8-9-24-36-2T or 8-9-8-24-36-2T or 9-8-8-24-36-2T - you know, change one parameter at a time, so I need to know which parameter is the most sensitive that I should change first.
 
William Gaatjes

William Gaatjes

Lifer
May 11, 2008
22,220
1,411
126
mtl77733 said:
Thanks, man!

Before I started this thread, I already tested 9-9-9-24-33-2T for 9 hours and it passed without any errors. I just want to see if I can improve the timings any further. So, if 8-8-8-24-36-2T is not working perfectly, I'm hoping someone could guide me as to what timings to try - maybe 8-8-9-24-36-2T or 8-9-8-24-36-2T or 9-8-8-24-36-2T - you know, change one parameter at a time, so I need to know which parameter is the most sensitive that I should change first.
Click to expand...

Good to see it does not have errors with more relaxed timings.
I started here reading about memory timings to get a more detailed understanding :
https://www.hardwaresecrets.com/understanding-ram-timings/
https://www.overclock.net/forum/18051-memory/410205-memory-timings-fully-explained.html
The overclockers forum should have some good threads about overclocking, the good thing is that even if you do not overclock, there is still valuable information.
 
LTC8K6

LTC8K6

Lifer
Mar 10, 2004
28,520
1,575
126
mtl77733 said:
The CPU is an AMD Phenom II X3 720 BE.

That PDF you linked to is only telling half the story. It's just a list of tested compatible memory but it does not say if the DDR3-1600 modules in the list will run without overclocking the CPU.

Please see the specs here...

https://www.asrock.com/mb/AMD/M3A790GXH128M/index.asp

In the memory section it says, "Supports DDR3 1600(OC)/1333/1066/800 non-ECC, un-buffered memory". Note the "OC" for DDR3 1600, which means the CPU must be overclocked in order to be able to use DDR3-1600.

Also, another problem is that my BIOS actually doesn't have exact 800 MHz as an option. There is 667 MHz and after that is 833 MHz. 833 MHz is too much so I can't use it even if I overclock.
Click to expand...
Every mobo lists the mem speeds that way. It doesn't necessarily mean you can't run the mem at those speeds.

You didn't mention 833 before, and that could certainly be too high.

However, the specs list 1600...which would be 800...I dunno...

Looking around the net, I see lots of people running their ram at 1600 with that CPU.
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
LTC8K6 said:
Every mobo lists the mem speeds that way. It doesn't necessarily mean you can't run the mem at those speeds.

You didn't mention 833 before, and that could certainly be too high.

However, the specs list 1600...which would be 800...I dunno...

Looking around the net, I see lots of people running their ram at 1600 with that CPU.
Click to expand...

But maybe not with the Asrock M3A790GXH/128M mobo. I'd be surprised to see them able to run 1600 RAM on this mobo without overclocking.
 
SlowBox

SlowBox

Member
Jul 4, 2018
80
5
16
mtl77733 said:
I bought this Ballistix DDR3-1600 memory here - http://www.crucial.com/usa/en/blt2kit4g3d1608dt1tx0 and I have a question. It doesn't work at 800 MHz 8-8-8-24-36-2T timings for my mobo (Asrock M3A790GXH/128M) because my mobo requires overclocking to be able to use DDR3-1600. So, without overclocking, my next choice is 666 MHz = DDR3-1333. Can I use this memory with timings of 8-8-8-24-36-2T for 666 MHz? If not, then what timings do you suggest for 666 MHz?

Also I have another question. This Ballistix memory runs at 1.5V but I can't select 1.5V in my BIOS. The only options that are close to 1.5V are 1.48V and 1.53V. Can I use 1.53V? It's not too much voltage - only 0.03V higher, right? Do you think it's too much? I hope not.

Thanks in advance for your much appreciated help!
Click to expand...

Another one, dude you can't cuss and call people the A word,.... nah jk :) oh the board ...

Ok those timings should work on a 1066 or 1333 ,, hmmmmmm maybe you have to up the DIMM voltage. Your good up to 1.65v so bump it to 1.600 and loosten the timings for testing purposes. let us know
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
SlowBox said:
Another one, dude you can't cuss and call people the A word,.... nah jk :) oh the board ...

Ok those timings should work on a 1066 or 1333 ,, hmmmmmm maybe you have to up the DIMM voltage. Your good up to 1.65v so bump it to 1.600 and loosten the timings for testing purposes. let us know
Click to expand...

Are you saying I'll be able to use 8-8-8-24-36-2T if I bump up the DIMM voltage to 1.6V? I thought the max voltage for my RAM is only 1.5V. Where did you see that it can go up to 1.65V?
 
Last edited:
cmdrdredd

cmdrdredd

Lifer
Dec 12, 2001
27,052
357
126
mtl77733 said:
We need to qualify how many passes for which we consider an error as fail. How many passes is enough? Because we can do 100 passes which will take a very long time and if an error occurs on the 100th pass then we consider it as fail. But on the other hand, if we do only 70 passes and no error happens and we're satisfied with that, then we consider it as success. So the 100 passes is fail but the 70 passes is success.

Come to think of it, we can never really be 100% certain of success unless we do a very high number of passes, say 10,000 passes, but do we have time for that? So, for practicality, we need to set a limit as to what is good enough or reliable enough, as nothing can really be 100% reliable in our imperfect world.

If I'm going to increase the timings a bit, which timing parameter should I start increasing first (CL or tRAS or tRCD or tRP or tRC, etc)? Which timing parameter is the most sensitive?
Click to expand...

I say 24hours continuous. Not unlike extreme CPU stability testing.
 
LTC8K6

LTC8K6

Lifer
Mar 10, 2004
28,520
1,575
126
SlowBox said:
Another one, dude you can't cuss and call people the A word,.... nah jk :) oh the board ...

Ok those timings should work on a 1066 or 1333 ,, hmmmmmm maybe you have to up the DIMM voltage. Your good up to 1.65v so bump it to 1.600 and loosten the timings for testing purposes. let us know
Click to expand...
If he's going to tweak stuff, he might as well overclock the chip and be done with it. :D
 
VirtualLarry

VirtualLarry

No Lifer
Aug 25, 2001
56,571
10,207
126
M

mtl77733

Member
Sep 12, 2013
33
0
66
According to the specs, the max voltage for my RAM is 1.5V. Do you think it's safe to use it at 1.6V, which is only 0.1V higher, or am I taking a big risk? FWIW, I remember I once had an OCZ Platinum RAM with max voltage of 1.8V and I used it at 2.0V for so many years and it never got damaged.

Also there's an option in the BIOS related to memory that is called "Channel Interleaving". The choices here are Disabled, Address Bits 6, Address Bits 12, Hash 1, and Hash 2. What is Channel Interleaving and what would determine the appropriate option to select for this?
 
Last edited:
William Gaatjes

William Gaatjes

Lifer
May 11, 2008
22,220
1,411
126
mtl77733 said:
According to the specs, that max voltage for my RAM is 1.5V. Do you think it's safe to use it at 1.6V, which is only 0.1V higher, or am I taking a big risk? FWIW, I remember I once had an OCZ Platinum RAM with max voltage of 1.8V and I used it at 2.0V for so many years and it never got damaged.

Also there's an option in the BIOS related to memory that is called "Channel Interleaving". The choices here are Disabled, Address Bits 6, Address Bits 12, Hash 1, and Hash 2. What is Channel Interleaving and what would determine the appropriate option to select for this?
Click to expand...

Jedec spec seems to be 1.8V absolute maximum for DDR3. But that can mean the memory survives a voltage glitch and says nothing about reliable operation when the voltage is 1.8V.
It is at your own risk.

Channel interleaving:
https://en.wikipedia.org/wiki/Interleaved_memory
The best setting depend on the memory chips used.
Try address bits 6 and 12 option and see if that makes a difference in a benchmark test.
That would be my starting point.
 
M

mtl77733

Member
Sep 12, 2013
33
0
66
William Gaatjes said:
Jedec spec seems to be 1.8V absolute maximum for DDR3. But that can mean the memory survives a voltage glitch and says nothing about reliable operation when the voltage is 1.8V.
It is at your own risk.

Channel interleaving:
https://en.wikipedia.org/wiki/Interleaved_memory
The best setting depend on the memory chips used.
Try address bits 6 and 12 option and see if that makes a difference in a benchmark test.
That would be my starting point.
Click to expand...

Thanks, man! I currently have it at Address Bits 6. What's a good RAM benchmark test that I can use?
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom