What's the best motherboard/CPU/RAM upgrade for me??

Memory tab:

vwsoyt.jpg


I was just downloading the "Superposition" benchmark from unigine - any good to use??

In respect of temps, what do I need to keep an eye on and which temps in particular? (sorry to keep asking and sorry this seems to have shifted from an "upgrade" post to an "overclocking" post!)
 
StevieP said:
I was just downloading the "Superposition" benchmark from unigine - any good to use??
Click to expand...
That's geared more to stressing the GPU.

A quick burst of prime95 should identify any problems for such a minor clock - but as mentioned it will be everyday usage that will be the true test.

StevieP said:
In respect of temps, what do I need to keep an eye on and which temps in particular? (sorry to keep asking and sorry this seems to have shifted from an "upgrade" post to an "overclocking" post!)
Click to expand...

No worries bud - not a problem.

Ideally download Coretemp and monitor the temps as you run the blend test - combination of CPU and memory (however your memory may be running under spec). It should be fine as all you've done is upped the multiplier.

StevieP said:
Memory tab:

vwsoyt.jpg
Click to expand...

Your memory is only running at 1333MHz - could you show me the SPD tab please.
 
StevieP said:
1r25u1.jpg
Click to expand...

Yep, you're running your memory under spec.

For speed, you could run the XMP within the BIOS - but then manually set the voltage to 1.55v (would ideally be 1.5V - but we're testing and it won't cause any harm).

This will set the Frequency to 1600MHz but under volt the memory to 1.55V - your memory should be fine with this.
 
Right :)

I took your advice and enabled XMP which seemed to then say that the RAM voltage was 1.65V (greyed out underneath the "Extreme Memory Profile (X.M.P)" line below but then overtyped "Auto" with 1.55 in the "DRAM Voltage" field (is this the right field?) - tried to exit and save and the PC died - then rebooted and died again almost immediately during boot!

avihhz.jpg


I then turned it on again and it's booted and within CPU-Z it now says the following on the Memory and SPD tabs:

28wmcgj.jpg
i6wo7n.jpg


Has this worked OK?? (I'm specifically looking at the "FSB:RAM" saying 1:6?? and more importantly, have I changed the correct settings!?!?!)
 
Yes, it's running at 1600MHz and from the BIOS it's at 1.55V.

The boot hiccup may have been the system just adjusting to the new settings.

If it becomes unstable bump it up to 1.6V - it's not a problem - but the ideal is 1.5V when clocking on the p67 - but not crucial at this stage of your CPUs/motherboards life (really don't worry if you have to up the voltage).

If you get cold boot issues - or when you run your stress tests you can stability issues bump up the memory voltage first.
 
Great, I'll do a bit of gaming tonight and see how it fares!

Just for my own knowledge, my understanding of increasing voltages is that you do it to provide more power to the component that is running at a higher speed - with the tradeoff being that the temperatures go up in line with the voltage - is this right?

Also, I assume I'm effectively "undervolting" the RAM as the "XMP-1600" profile asks for 1.65V? If this is right, why would 1.5V be ideal and why can't I set it to 1.5V? (is it because the RAM needs that extra voltage to run at 1600MHz but 1.65V would overheat them??)

Sorry to ask for the lessons but I prefer to understand why I'm doing what I'm doing rather than just being happy if it works!

Thanks again....
 
StevieP said:
Just for my own knowledge, my understanding of increasing voltages is that you do it to provide more power to the component that is running at a higher speed - with the tradeoff being that the temperatures go up in line with the voltage - is this right?

Also, I assume I'm effectively "undervolting" the RAM as the "XMP-1600" profile asks for 1.65V? If this is right, why would 1.5V be ideal and why can't I set it to 1.5V? (is it because the RAM needs that extra voltage to run at 1600MHz but 1.65V would overheat them??)
Click to expand...

When we all first starting overclocking the Sandybridge chips, many moons ago now, INTEL'S advice stated that 1.59v was the maximum voltage the memory controller could supposedly handle (especially when overclocking) - but ideally you should get as close 1.5V as possible.

So that's what we all dutifully did, with much pride, undervolting our 1600MHz (1.65v) approved sticks at 1.5v/1.55v. However, i've come across countless rigs/threads where people have been running their memory at 1.65V the whole time as they set the XMP and didn't change the voltage.

The reality is it seems the memory controller was more robust than INTEL anticipated - or they deliberately overstated/exaggerated the issue to stop overclockers abusing the MC even more - i.e. going over 1.65v (which we would have if we were told 1.65V was safe - we're idiots)

And it's technically a problem that would have appeared over time, reducing the life of the CPU, but your's is already 5+ years old - i think it will be able to handle 1.6V if we have to feed it that much :)
 
Thanks for the info!

Not the best experience so far though - jumped on The Division with a mate and within 90 seconds the PC crashed!

First step to increase the voltage? If so, in what increment?
 
It may be the clock - not the memory. We've changed 2 things - which in retrospect probably wasn't wise as it could be one of 2 things causing the instability (my bad).

For now - set the memory back to auto - resetting it back to 1333Mhz.

While you're in the BIOS check that the multiplier is still set to 40 - and then save and exit.

If it bulks in Division after that then the clock you applied is the problem - which will mean applying manual voltages.

Do the above - have a play and report back when you're done.
 
I would not bother overclocking the ram at all. If it has XMP set that to profile 1 and call it good. If not leave it at stock as overclocking ram is a minefield in comparison to cpus/gpus. The speed difference isn't going to be noticeable anyway.

Just work on the cpu Stevie :), with voltage you can increase it by using numpad + usually, this will increase it in really small increments. But as a general rule upto 1.45 is agreed to be fine for 24/7 operation.
 
harry5522 said:
I would not bother overclocking the ram at all. If it has XMP set that to profile 1 and call it good. If not leave it at stock as overclocking ram is a minefield in comparison to cpus/gpus. The speed difference isn't going to be noticeable anyway.

Just work on the cpu Stevie :), with voltage you can increase it by using numpad + usually, this will increase it in really small increments. But as a general rule upto 1.45 is agreed to be fine for 24/7 operation.
Click to expand...

I think you've scan read the posts again harry (read back about 10 posts) :D - we only set the xmp (no memory overclock) and then adjusted the Dram voltage. If he sets it to profile 1 it will bump it up to 1600MHz again and boost the Dram to 1.65V (don't do this stevie)

We weren't overclocking the memory - it was underclocked before at 1333MHz instead of its XMP setting of 1600 MHz. i was testing to see if it could run at the XMP but at the INTEL recommended sub 1.59v - but this can be tested later.

The clock was a setting Micky recommended - i think he simply suggested to Stevie to up the multiplier by 40 and test - sometimes works.

I had a hunch his memory was running under stock (it was) so we applied the XMP but undervolted as is the norm - all Sandybridges i've setup have run at 1600MHz with no issues so suspect the clock will need more juice.

Or, the combination of 4GHz plus the XMP is too much without some voltage tweaks...

I've normally manually set my memory to the XMP settings and then applied the CPU clock - but all CPUs/boards are different...

TBH, this isn't how i clock CPUs - i like to make sure everything is stable at stock (memory included) - and then pick a clock and adjust voltages, BIOS teaks, as necessary.
 
Last edited:
Um yeah.... that didn't go so well! :p

Put the RAM back to auto and just tried running Prime95 with the CPU at 4GHz - immediately froze my PC and crashed to the "Something went wrong, we are collating information" page! (came up with the error message "CLOCK_WATCHDOG_TIMEOUT" if that helps!)

Just for my own peace of mind I've put everything back to stock and am running Prime95 again - so far it's been fine for 30 mins+ with the 4 cores hovering between 49-53 degrees! (but as I have no idea what I'm doing - I don't even know if these temps are about right!)

Hmmmm, just noticed something strange - CPU-Z says its running at 3399.21 MHz - since I put it back to "Auto", shouldn't it be running at 3.3GHz and not 3.4GHz??
 
StevieP said:
Um yeah.... that didn't go so well! :p

Put the RAM back to auto and just tried running Prime95 with the CPU at 4GHz - immediately froze my PC and crashed to the "Something went wrong, we are collating information" page! (came up with the error message "CLOCK_WATCHDOG_TIMEOUT" if that helps!)

Just for my own peace of mind I've put everything back to stock and am running Prime95 again - so far it's been fine for 30 mins+ with the 4 cores hovering between 49-51 degrees!
Click to expand...

Yes, i'm not fan of this overclock method - it's too hit and miss. It was worth a try - but it's not workong for you.

This isn't how i clock CPUs - i like to make sure everything is stable at stock (memory included) - and then pick a clock and adjust voltages, BIOS teaks, as necessary.

I would set the your multiplier back to auto/34 save and exit - which you've done.

Then if you want we can get your memory set at the XMP and this time manually set 1.59v (no point playing with 1.55) - and then test.

Once we've got your memory stable we can look at clocking your CPU properly - 4.2GHz probably a safe middle ground but applying the vCore manually and working our way down incrementally until stable at the lowest voltage possible.

Thoughts?
 
Last edited:
StevieP said:
Just for my own peace of mind I've put everything back to stock and am running Prime95 again - so far it's been fine for 30 mins+ with the 4 cores hovering between 49-53 degrees! (but as I have no idea what I'm doing - I don't even know if these temps are about right!)

Hmmmm, just noticed something strange - CPU-Z says its running at 3399.21 MHz - since I put it back to "Auto", shouldn't it be running at 3.3GHz and not 3.4GHz??
Click to expand...
Your temps are fine - great even.

And it is running at 3.4GHz - all bar 0.79Hz (normal) :)
 
Sounds good - although before we start have you any idea why the core speed reported within CPU-Z would change?

I've checked the BIOS and it's definitely set to "Auto" but it was showing the multiplier as 34 and now it's showing as 16!?!?!? (and seems to be flickering back an forth - shouldn't it just say 33??)

k2y51t.jpg
nqdypt.jpg


These 2 screenshots were taken only second apart!
 
To be fair having just seen the words Prime 95 i allow myself to skim read, P95 is not a good tool to be using. It puts an unatural strain on the cpu that it will never see during gaming loads. you really should just test it's stability via playing the games you play.

The speed changes because of turbo boost and power saving modes will throttle the cpu down really slow when its not being used. :) I forget the exact name of it on that platform but i think it might be called c1e that causes it to drop to those speeds and turbo boost that brings it up to the high speed.
 
StevieP said:
Sounds good - although before we start have you any idea why the core speed reported within CPU-Z would change?

I've checked the BIOS and it's definitely set to "Auto" but it was showing the multiplier as 34 and now it's showing as 16!?!?!? (and seems to be flickering back an forth - shouldn't it just say 33??)
Click to expand...
It's speedstep - power saving feature which drops the Core speed and voltage, respectively, when the CPU is idle. Saves energy.
 
You must log in or register to reply here.
Back
Top Bottom