1600MHz ram - highest stable NB frequency for M5A99X EVO R2.0 ?

[Silver_Pharaoh]

Use hyper Pi to stress the NB. Either that or Minecraft.
If you dart around the map in Minecraft then your NB is unstable.

For reference, I can go as high as 2450Mhz NB before no amount of voltage gets it stable.

VDDA, I have found keeps the Vcore fluctuations on the higher end. My Vcore bounces between 1.42 and 1.46 under load. With VDDA set to 2.8 volts, it stayed mostly at 1.46. With VDDA set to auto, Vcore sat at 1.42 more often. YMMV, this might just be my board acting funny.

CPU/NB is the NB in the CPU.... AMD has 2 NB's. One one the CPU and one on the motherboard.
CPU/NB voltage would be the one you want here - that's the CPUs NB.
NB voltage is the motherboard's voltage. For higher FSB and such IIRC.
NB 1.8 would probably be "NB termination voltage" That's my best guess as I have a similar option on my Gigabyte board. I don't touch that voltage.

 

[Mandrake4565]

Use hyper Pi to stress the NB. Either that or Minecraft.
If you dart around the map in Minecraft then your NB is unstable.

For reference, I can go as high as 2450Mhz NB before no amount of voltage gets it stable.

VDDA, I have found keeps the Vcore fluctuations on the higher end. My Vcore bounces between 1.42 and 1.46 under load. With VDDA set to 2.8 volts, it stayed mostly at 1.46. With VDDA set to auto, Vcore sat at 1.42 more often. YMMV, this might just be my board acting funny.

CPU/NB is the NB in the CPU.... AMD has 2 NB's. One one the CPU and one on the motherboard.
CPU/NB voltage would be the one you want here - that's the CPUs NB.
NB voltage is the motherboard's voltage. For higher FSB and such IIRC.
NB 1.8 would probably be "NB termination voltage" That's my best guess as I have a similar option on my Gigabyte board. I don't touch that voltage.

Hyper pi isn't going to stress it long enough to determine whether it's stable or not SP, yeah if it passes it's stable for the bench but still may not be for daily Oc's.

There is a big difference between the VDDA voltage on the Gigabyte board and the Asus board. I do not recall seeing any proof raising the VDDA voltage on the Asus board help stabilize an Oc where as this has been some instances shown where it does help on the Giga board. I believe the reason is because the Asus VRM is digital and the Giga is analog.

SP is correct about the voltages.

Additionally if you want to do some reading here's some pretty good guides.
Asus Crosshair V Formula overclocking guide, this will be helpful with some of the features on your board. You do not have all of the available adjustable fields of the CHV but the ones you do have you may get a better idea what they do.

This one is directly from AMD, it's the Performance tuning guide for the Fx. View attachment AMD_FX_Performance_Tuning_Guide.pdf

 

[Silver_Pharaoh]

Hyper pi isn't going to stress it long enough to determine whether it's stable or not SP, yeah if it passes it's stable for the bench but still may not be for daily Oc's.

There is a big difference between the VDDA voltage on the Gigabyte board and the Asus board. I do not recall seeing any proof raising the VDDA voltage on the Asus board help stabilize an Oc where as this has been some instances shown where it does help on the Giga board. I believe the reason is because the Asus VRM is digital and the Giga is analog.

SP is correct about the voltages.

Additionally if you want to do some reading here's some pretty good guides.
Asus Crosshair V Formula overclocking guide, this will be helpful with some of the features on your board. You do not have all of the available adjustable fields of the CHV but the ones you do have you may get a better idea what they do.

This one is directly from AMD, it's the Performance tuning guide for the Fx. View attachment 170613

That's true, but I suppose you could run Hyper pi a few times.
I think Prime95 Large FFT will stress the NB too no?

Good point on the VDDA, forgot that Giga still used analog VRMs

 

[Mandrake4565]

I always just use Prime Blend, it works for me. If I can pass 2 hours my pc can take anything I throw at it.

 

[Kenrou]

VDDA, I have found keeps the Vcore fluctuations on the higher end. My Vcore bounces between 1.42 and 1.46 under load. With VDDA set to 2.8 volts, it stayed mostly at 1.46. With VDDA set to auto, Vcore sat at 1.42 more often. YMMV, this might just be my board acting funny.

Agreed, in my case VDDA at 2.5v results in IBT being faster by ~5gflops but failing and if i bump it to 2.6v its slower but passes without problems (like SP said might be the board being funny).

 

[Silver_Pharaoh]

Agreed, in my case VDDA at 2.5v results in IBT being faster by ~5gflops but failing and if i bump it to 2.6v its slower but passes without problems (like SP said might be the board being funny).

Really? I dunno, either VDDA does something unique to all boards, or your board was acting wayyyyy funny for that to happen

 

[Sologuy]

Thanks for the tips guys, much appreciated. I"m still going to stick with IBT for stress testing as it has served my well since its incarnation some years ago (2009?) albeit with updates. I've seen systems run p95 for many hrs like 24 or more and yet still crash in some games, the source of the crashes was not gpu related either. Could it have been the mobo? we will probably never know...

Majorgeeks.com have a good description of IBT here

As a side note, I'm only really interested in stability for the games I play on this system as it doesn't get stressed enough on all cores 24/7 to justify 100% stress testing on constant load.. waste of electricity imo..

There was also a BIG discussion last year on this forum as to why some OC's don't like p95..


Bottom line: as long as its stable enough for my uses.. that's all that matters..

 

[Kenrou]

Really? I dunno, either VDDA does something unique to all boards, or your board was acting wayyyyy funny for that to happen

I did ask here and in other forums if this is normal in Sabers, no one could give me a straight answer, My LLC has weird settings as well. But point being that VDDA stabilizes my high-ish OC.

 

[Kenrou]

Thanks for the tips guys, much appreciated. I"m still going to stick with IBT for stress testing as it has served my well since its incarnation some years ago (2009?) albeit with updates. I've seen systems run p95 for many hrs like 24 or more and yet still crash in some games, the source of the crashes was not gpu related either. Could it have been the mobo? we will probably never know...

Majorgeeks.com have a good description of IBT here

As a side note, I'm only really interested in stability for the games I play on this system as it doesn't get stressed enough on all cores 24/7 to justify 100% stress testing on constant load.. waste of electricity imo..

There was also a BIG discussion last year on this forum as to why some OC's don't like p95..


Bottom line: as long as its stable enough for my uses.. that's all that matters..

If you're just looking for stability for games the standard test in IBT will suffice, 1gb uses less voltage and produces less heat but as most will say here the system will not be completely stable for say folding. Will be ok for games and everyday use on most systems though.

 

[Sologuy]

If you're just looking for stability for games the standard test in IBT will suffice, 1gb uses less voltage and produces less heat but as most will say here the system will not be completely stable for say folding. Will be ok for games and everyday use on most systems though.

Yes, that's what I believe too. However running IBT on 'very high' or 'maximum' is the best way to stress test for upcoming future games that should be 64 bit based because current gen of consoles (which devs will target 1st priority) use 64 bit systems up to 8GB capacity. Although something like 'Batman Arkham knight' is suppose to use up to 12GB for PC platform!

 

[Sologuy]

For reference, I can go as high as 2450Mhz NB before no amount of voltage gets it stable.

Btw, how much NB voltage do you need for that stability & what capacity of RAM? can't tell by looking at your signature. I'm only asking because you & I have same cpu model.

Thanks.

 

[PolRoger]

So 2600 is default for FX chips? ok, so I"ve been ripped off by "auto" settings in bios for this value.... hmmmm.... can't trust Asus huh?
Anyway, here are some screenshots from IBT stress testing with current settings experimenting as from OP..

View attachment 170542

However... starting the stress testing again for 2nd time produces this...

View attachment 170543

Not sure what was happening in 1st run.. GFlops are lower value.. makes me think IBT is not accurate at measuring GFlops reliably? and if that's the case, then how can we be sure its actually stress testing properly without having to resort to something like p95 and run it for hours, something I'll have to consider I suppose. For my usage of this machine, its good enough for gaming simply because no game I've ever heard of stresses all cpu cores 100% for significant periods of time anyway...

Your IBT results are "bugged"... Even though it says you have passed... You have a bad result: -1.#IND00e+000 with low GFLOPS.


Your results should not start with or have any -1 numbers... See below:




Edit: Link to IBT AVX: http://www.overclock.net/attachments/13202

 

[Silver_Pharaoh]

Btw, how much NB voltage do you need for that stability & what capacity of RAM? can't tell by looking at your signature. I'm only asking because you & I have same cpu model.

Thanks.

My NB was set to 1.3volts.
My CPU NB was about 1.35, never higher than 1.45.

Let me double check. Also note that your Sabertooth board is superior to my GA970-UD3

EDIT: Checked the BIOS but I didn;t save the profile with the NB set to 2450. I recall having the NB higher than 1.2 so I edited that info.

 

[Kenrou]

Each board is different, mine is at 1.25v at 2600mhz, might not pass for you even if you have the same one.

 

[Sologuy]

Your IBT results are "bugged"... Even though it says you have passed... You have a bad result: -1.#IND00e+000 with low GFLOPS.


Your results should not start with or have any -1 numbers... See below:

View attachment 170744


Edit: Link to IBT AVX: http://www.overclock.net/attachments/13202

Thanks for the observation but according to the author of this software, those using AMD systems are not mean't to run at all... > IntelBurnTest 2.54

So imo, I'm lucky I have other tools to check if cpu is under full load during testing...

My NB was set to 1.3volts.
My CPU NB was about 1.35, never higher than 1.45.

Let me double check. Also note that your Sabertooth board is superior to my GA970-UD3

EDIT: Checked the BIOS but I didn;t save the profile with the NB set to 2450. I recall having the NB higher than 1.2 so I edited that info.

Thanks, I'll take those volts into consideration when manually entering them next time I boot that system up but for the record, I don't have sabertooth board.. just plain old 990X chipset mobo... lol..

 

[Silver_Pharaoh]

Thanks, I'll take those volts into consideration when manually entering them next time I boot that system up but for the record, I don't have sabertooth board.. just plain old 990X chipset mobo... lol..

My bad, I was looking at Kenrou's sig.
Make sure you keep the NB and VRM heatsink's cooled with a fan too

 

[PolRoger]

Thanks for the observation but according to the author of this software, those using AMD systems are not mean't to run at all... > IntelBurnTest 2.54

So imo, I'm lucky I have other tools to check if cpu is under full load during testing...

IBT works fine on FX if your settings are dialed in correctly...

http://www.overclock.net/t/1519033/5-ghz-24-7-oc-club/360#post_23465516

It is used in this thread here over on OCN where there are many AMD FX cpu(s) submitted... Either drop you overclock or further tweak your settings... You should be able to make it work.

http://www.overclock.net/t/1519033/5-ghz-24-7-oc-club

 

[Sologuy]

IBT works fine on FX if your settings are dialed in correctly...

http://www.overclock.net/t/1519033/5-ghz-24-7-oc-club/360#post_23465516

It is used in this thread here over on OCN where there are many AMD FX cpu(s) submitted... Either drop you overclock or further tweak your settings... You should be able to make it work.

http://www.overclock.net/t/1519033/5-ghz-24-7-oc-club

Funny thing is it works on my FM2+ platform albeit with low GFlops for my A10-7850K and have 'proper' result... I have never been able to get it to work properly on this AM3+ platform even at stock! Imo, this software came out in mid 2012 so any AMD hardware released after this time in particular will never give accurate results. But one thing is certain, it most definitely puts the system under full load on all cores, task manager can verify that quite easily. As far as I'm concerned if you don't get a 'fail' message after 10 runs or more, then its fine for a quick stability test. Besides that, the author of this software clearly states it's not designed for AMD platforms.