[Rokabud]

Hi, I'm getting ready to overclock for the first time and after hearing the importance of voltages in achieving a stable OC, I have a question.

I found multiple guides with recommendations for voltages, but I'd like to know if there's a place I can find the "official" maximum suggested voltages for RAM (and other hardware.) I checked the Intel design guide, but it doesn't appear to be in there. Would this information be provided by the motherboard manufacturer or Intel? Or do both have separate voltage limits?

Thanks!

 

[appleg33k85]

Its going to be specified by the memory manufacture - unless you're doing some heavy memory overclocking you shouldn't need more than the manufacture recommended. Most memory will be 1.5v (low voltage) or 1.65 (normal)

 

[\(ao? a? c>Sa? )/ cutie pie]

If your RAM has no heatspreaders, or if they are slide-off, here's what you do:

1. Under good lighting, try to make out the numbers / markings on the chips themselves.

2. Go to your favorite search engine and look these up. Usually, but not always, the longest is the actual part number. You can also try a datasheet search for this.

3. Find and download a datasheet for the memory. You will find the max. voltage somewhere in here, usually rated in 'Vdd'. It may be 'Vcc', as though not correct, this term has been in use for so long plenty of people just use it as shorthand for 'supply voltage'.

You may very well find these ratings under a header titled "Absolute Maximum Specifications" or something similar. If you want to back off a few fractions of a volt, or even push it a bit, that's up to you. Personally I'd advise staying at least 1/20th of a volt away from this.

 

[Rokabud]

Quote:

Originally Posted by appleg33k85;13118298
Its going to be specified by the memory manufacture - unless you're doing some heavy memory overclocking you shouldn't need more than the manufacture recommended. Most memory will be 1.5v (low voltage) or 1.65 (normal)

You mean the rated voltage? I was wondering where to find the maximum (i.e. don't exceed 1.xxV or you'll be causing damage.)

Also, I thought of a few more RAM questions:

-How important is it to have undervolted memory? Does it just allow you to OC it higher?

-I heard that all RAM has a maximum stock frequency (1600 something?) and that if a set of RAM lists higher specs, it means that its rated to overclock to those speeds, but it comes at the maximum stock frequency.

-If that's true, are the timings listed in the specs the stock ones or are they loosened some to account for the frequency increase that the RAM is rated to achieve?

-Why does memory come in set frequencies (i.e 1333, 1600, etc.)? Why isn't it like a CPU where you can have it range in speed anywhere between, say, 3 and 4GHz. Why does RAM have those "steps" in speeds?

Thanks.

 

[Rokabud]

bump, anyone able to clear any of those questions up?

Thanks

 

[Mr.Eiht]

I read @G.Skill forum that they officially support up to 1,65v for DDR3 (not he low volt ones). So you can use 1,65v without voiding your warrenty. They also wrote there that you could use higher volts but they don`t support that - what ever "support" means.
(They should have said : "If you use more than 1,65v and you kill them: Not our fault")
So stay below 1,65v or @1,65v and you are fine.
Edit:
http://www.gskill.us/forum/showthrea...highlight=volt

 

[MoCo Man]

No matter what the RAM specs are, Intel says to NEVER exceed 1.65V on your ram. It will damage your CPU.

Newer ram never needs to go above that anyway.

Go to the INTEL site to get the data sheet for your processor. The voltage range they state for your CPU will be what they recommend. Check the forums here to see that others may excede those, it up to you, be aware that higher voltages than INTEL recommends will void warranty and may reduce the life of your components.

 

[UsedPaperclip]

I can't remember where I saw it but intel support started the max safe ram voltage when raising the voltage on the vcore is 1.5. I guess they said raising it past 1.5 can risk damaging the memory controller while the vcore is raised. I don't have anything to back that up but you may be able to search for something along those lines.

 

[MoCo Man]

Quote:

Originally Posted by Rokabud You mean the rated voltage? I was wondering where to find the maximum (i.e. don't exceed 1.xxV or you'll be causing damage.) Also, I thought of a few more RAM questions: -How important is it to have undervolted memory? Does it just allow you to OC it higher? -I heard that all RAM has a maximum stock frequency (1600 something?) and that if a set of RAM lists higher specs, it means that its rated to overclock to those speeds, but it comes at the maximum stock frequency. -If that's true, are the timings listed in the specs the stock ones or are they loosened some to account for the frequency increase that the RAM is rated to achieve? -Why does memory come in set frequencies (i.e 1333, 1600, etc.)? Why isn't it like a CPU where you can have it range in speed anywhere between, say, 3 and 4GHz. Why does RAM have those "steps" in speeds? Thanks.

There are great links to guides to overclock any of the common processors. Make sure your read these over very carefully. Join a forum on your processor, ask alot of questions to the folks there.

 

[Mr.Eiht]

What you mean by undervolt RAM?
There are RAMs that need lower volts (low voltage RAM).
Anyway to run the memory @higher clocks you need more volts.

Quote:

Originally Posted by xxbassplayerxx Also, keep in mind that any DDR3 above 1066 is considered overclocked.

This.

 

[Rokabud]

Quote:

Originally Posted by Mr.Eiht What you mean by undervolt RAM? There are RAMs that need lower volts (low voltage RAM). Anyway to run the memory @higher clocks you need more volts. This.

So 1066 is the highest stock frequency for RAM?

I was looking on the intel site for the suggested RAM voltages and couldn't find it. This was the closest info I could find:

http://ark.intel.com/Product.aspx?id=52210

Quote:

There are great links to guides to overclock any of the common processors. Make sure your read these over very carefully. Join a forum on your processor, ask alot of questions to the folks there.

Yes, but I've noticed that in most of the Sandy Bridge overclocking guides I've read, the RAM doesn't receive much attention.

In older overclocking guides, there was a lot of information about "CPU:RAM ratios", but in the Sandy Bridge guides, it seems that the CPU is overclocked generally independent of RAM.

Are CPU:RAM ratios no longer important on Sandy Bridge systems? Since the FSB on Sandy Bridge processors is locked, I'd assume you'd have to OC both components through their respective multipliers. So what are important specs to look for in RAM for a Sandy Bridge system, then? Obviously the voltage limitations, but anything else?

I've heard that it's important to get RAM that has a frequency that can be achieved evenly through the FSB, and since SB's FSB is locked at 100, frequencies that can be evenly divided by 100. Is there any truth to that?

 

[FtW 420]

Don't know too much sandy bridge specific info yet, but have read before that ddr3 memory according to JEDEC standards must be able to withstand up to 1.85V before incurring damage to the memory modules (nothing about cpu & don't remember the source). On x58 I've run memory up to 1.9V for short periods.

Sandy is supposed to be rated for 1.5V memory, & read that it can be safely run up to 1.58V (again no source, read that here on OCN somewhere). On my 2500k rig I usually run the memory at ~1.65V, gone up to 1.74V for short periods but haven't used it long enough to say if there is any damage to anything, nothing noticeable yet at least.
With SB you are pretty much stuck to overclocking memory with the memory divider, so 1066, 1333, 1600, 1866 & 2133. You can adjust it a small amount with the bclk, but memory options are more limited than previous chipset generations.

 

[FtW 420]

too many double post from me today, gotta cut back on the coffee.

 

[Mr.Eiht]

Rokabud this quote from me came from
xxbassplayerxx
a respectable member of the OCN society.
The stock speeds of your RAM can be considered as o.c`ed if you like it this way.
Since the producer of the RAM sells it @this (e.g. 1600MHz) stock speed the producers have tested that they are stable

 

[Rokabud]

Quote:

Originally Posted by FtW 420 Don't know too much sandy bridge specific info yet, but have read before that ddr3 memory according to JEDEC standards must be able to withstand up to 1.85V before incurring damage to the memory modules (nothing about cpu & don't remember the source). On x58 I've run memory up to 1.9V for short periods. Sandy is supposed to be rated for 1.5V memory, & read that it can be safely run up to 1.58V (again no source, read that here on OCN somewhere). On my 2500k rig I usually run the memory at ~1.65V, gone up to 1.74V for short periods but haven't used it long enough to say if there is any damage to anything, nothing noticeable yet at least. With SB you are pretty much stuck to overclocking memory with the memory divider, so 1066, 1333, 1600, 1866 & 2133. You can adjust it a small amount with the bclk, but memory options are more limited than previous chipset generations.

Thanks

This actually relates to a question I posted in another thread:

Quote:

Hmm this has me even more confused. I thought both the CPU and RAM operated off the FSB, and achieved different speeds via their independent multipliers. I thought the FSB=BLCK. Is this not true? I thought the base clock speed of the CPU was the speed at which both the CPU and RAM operated before multiplier. In some overclocking guides I've read they explain that its best to achieve the goal speed by having a higher FSB and a lower multiplier; so that while your CPU would operate at the same speed if you used a lower FSB and higher multiplier, it would be able to communicate with the RAM faster because of the higher FSB. What's happened since to change that?

Does this mean the only three factors involved with RAM overclocking (besides temperature) is the voltage, frequency and timings? You don't have to compare it to the CPU whatsoever? There are no specific ratios of CPU to RAM speed that would provide better performance? No "sweet spots"?

And about the locked BLCK on SB systems, would increasing the BLCK on Sandy Bridge rather than the multiplier still yield better results if it were possible? Why was the BLCK locked in the first place then? Would it not be better to overclock through that then the multiplier?

Thanks.

 

[Blameless]

"Official" voltages, "safe" voltages, required voltages, and reasonable voltages may all be different numbers.

Officially, no DDR3 ICs are rated above 1.5v for their nominal voltage, and all are required to handle 1.575v long term. However, all DDR3 ICs are required to withstand 1.975v for (very) short periods without permanent damage.

In practicality, safe voltages will differ, and this will depend on many factors. Some early DDR3 ICs will handle 2v all day and night without issue. Other ICs will fail at 1.6v, if used for prolonged periods.

Much DDR3, especially those that are rated for 1.65v, or that are otherwise outside of JEDEC spec, is essentially binned and warranted for a certain OC by the manufacturer.

Quote:

Originally Posted by Rokabud And about the locked BLCK on SB systems, would increasing the BLCK on Sandy Bridge rather than the multiplier still yield better results if it were possible? Why was the BLCK locked in the first place then? Would it not be better to overclock through that then the multiplier?

No.

BCLK, in and of itself is irrelevant. It's not a front-side bus, or any bus at all. It's just a reference for multipliers to work off.

This has been standard for AMD since the first Athlon 64s and for Intels since the original core i7.

If CPU, uncore, and memory speeds are the same, it does not matter what BCLK is used to reach those speeds, performance will remain the same.

BCLK was "locked" because it's simpler to tie all the buses to it (allowing asynchronous clocks requires multiple clock generators or the addition of many PLLs to separate clock domains), and because it allows Intel to charge a modest premium for CPUs capable of significant OCing. It also prevents market cannibalization because all the K series CPUs are fairly high-end ones. The days of buying 50-100 dollar CPUs and matching the performance of a 300+ dollar chip are over.

 

[Rokabud]

Quote:

Originally Posted by Mr.Eiht Rokabud this quote from me came from xxbassplayerxx a respectable member of the OCN society. The stock speeds of your RAM can be considered as o.c`ed if you like it this way. Since the producer of the RAM sells it @this (e.g. 1600MHz) stock speed the producers have tested that they are stable

That's what it seems like to me. Looking at the stock specs of RAM, as the frequency increases, the timings loosen and the voltages increase. It's like the manufacturer is just overclocking RAM and not making better or worse models.

And that's why I'm so confused about what to look for.

Quote:

"Official" voltages, "safe" voltages, required voltages, and reasonable voltages may all be different numbers. Officially, no DDR3 ICs are rated above 1.5v for their nominal voltage, and all are required to handle 1.575v long term. However, all DDR3 ICs are required to withstand 1.975v for (very) short periods without permanent damage. In practicality, safe voltages will differ, and this will depend on many factors. Some early DDR3 ICs will handle 2v all day and night without issue. Other ICs will fail at 1.6v, if used for prolonged periods. Much DDR3, especially those that are rated for 1.65v, or that are otherwise outside of JEDEC spec, is essentially binned and warranted for a certain OC by the manufacturer.

My RAM's specs (http://www.newegg.com/Product/Produc...82E16820231303) list 1.65V. So far from what I've heard that's pretty high for use on a Sandy Bridge system. I'd like to be able to OC the RAM, if only for the experience, but it looks like with what I have now I have no headroom.

 

[Rokabud]

Quote:

Originally Posted by Blameless No. BCLK, in and of itself is irrelevant. It's not a front-side bus, or any bus at all. It's just a reference for multipliers to work off. This has been standard for AMD since the first Athlon 64s and for Intels since the original core i7. If CPU, uncore, and memory speeds are the same, it does not matter what BCLK is used to reach those speeds, performance will remain the same. BCLK was "locked" because it's simpler to tie all the buses to it (allowing asynchronous clocks requires multiple clock generators or the addition of many PLLs to separate clock domains), and because it allows Intel to charge a modest premium for CPUs capable of significant OCing. It also prevents market cannibalization because all the K series CPUs are fairly high-end ones. The days of buying 50-100 dollar CPUs and matching the performance of a 300+ dollar chip are over.

Ohh I see, so the BLCK is just one factor used to achieve an OC, it doesn't hold any significance unless coupled with a multiplier to yield CPU speed?

Then what is the FSB? Where has it gone? Is it still manipulated when overclocking?

+rep everyone sorry if any of this sounds stupid

 

[FtW 420]

The FSB = baseclock (bclk), it is much different now in the Sandy Bridge systems than previous intel systems. Before when the cpu multi was maxed out overclocking was done by increasing the bclk, with SB the bclk is basically locked to 100Mhz & the overclocking is done by the multi.
You can still increase/decrease the bclk from 100 to change the memory & clockspeed, but is much more limited (maybe 10 either way max, 90 - 110), I've changed mine from 100 to 104.3 for some benching before, but don't really know yet if there are any dangers in leaving it overclocked for more than for short term testing.

 

[Blameless]

Quote:

Originally Posted by Rokabud Ohh I see, so the BLCK is just one factor used to achieve an OC, it doesn't hold any significance unless coupled with a multiplier to yield CPU speed? Then what is the FSB? Where has it gone? Is it still manipulated when overclocking? +rep everyone sorry if any of this sounds stupid

CPUs prior to the integration of the IMC use the FSB to connect the chipset to the CPU. Since everything was connected to the chipset (memory, AGP/PCI/PCI-E/DMI buses, other CPUs, all of it), the performance of this bus was often the determining factor in overall system performance.

There was obviously a clock generator that set the reference/base clock, but since the FSB was directly linked to it with a fixed ratio, there was almost never a reason to distinguish between the two, so FSB and reference clock became synonymous.

In modern CPUs, there is no front side bus. The memory connects directly to an on-die memory controller, and a separate bus (hypertransport, QPI, or DMI) connects the much simplified chipset to the CPU. On the newest CPUs, even the primary PCI-E bus is integrated. There is still a reference clock, but since it's not tied to any single bus, calling it the FSB is a fairly serious misnomer.

So we've gone from one pipe carrying all data, to different paths for most everything. The performance of the chipset to CPU connection has become much less of an issue now that memory traffic is independent of it.