This topic has been locked
Dr. Danger Nov 26, 2022 @ 11:48am
I managed to Undervolt my DDR3 RAM to 1.2v stable!
Yep.
I am amazed, it passed Memtest 3 times if I remember correctly.
RAM Module is Kingston KVR 1333mhz 1.5v.
CPU is i7 2600k and RAM is Dual Channel 4+4(8gb).
Motherboard is Lga 1155 H61 Chipset (Gigabyte)
Power Supply - Corsair CX 450m
Btw, undervolted all Chipsets and CPU on Mobo without any stability issues too, this is INSANE!

I was able to get 0 errors at 1.140v in a 2min test, but at 1.120v it gives some errors, like 60 total in 1min and at 1.100 it gives 65000+ or whatever is the max count for Memtest in just a few seconds. Now I am wondering should I try to reach stable 1.140 or 1.160 for even lower temps, higher stability, longer lifespan of modules.

I just built this PC, switched from DDR2 to DDR3 config and it's literally night and day difference.
GPU is currently HD 5870 1GB (very limiting factor) GTA 5 flies on this machine, as long as VRAM is within a GPU's limits. Around New Year I will upgrade that too probably, to maybe GTX 750ti, GTX 950/960 or maybe GT 1030.

Anyways, GPU is not my focus here, so anyone had experience like this? Is this a really well made module or what? Maybe they just took 1600mhz modules, named them 1333, limited their speed and that would explain why it runs at such a low Voltage without any issues...
Last edited by Dr. Danger; Nov 26, 2022 @ 11:48am

Something went wrong while displaying this content. Refresh

Error Reference: Community_9734361_
Loading CSS chunk 7561 failed.
(error: https://community.cloudflare.steamstatic.com/public/css/applications/community/communityawardsapp.css?contenthash=789dd1fbdb6c6b5c773d)
< 1 2 3 4 5 6 ... 10 >
Showing 1-15 of 137 comments
76561198343548661 Nov 26, 2022 @ 12:12pm 
Why ? I m hearing such a thing for the first time in my life .

My 10 year old laptop has DDR3 1333MHz RAM . The CPU/GPU temp tend to skyrocket but i have never had issues with the RAM . Some silicon can withstand very high temps . My laptop GT 520MX doesnt throttle even at 99C . But i m not that stupid and have underclocked it .
Dr. Danger Nov 26, 2022 @ 1:30pm 
Originally posted by smallcat:
Why ? I m hearing such a thing for the first time in my life .

My 10 year old laptop has DDR3 1333MHz RAM . The CPU/GPU temp tend to skyrocket but i have never had issues with the RAM . Some silicon can withstand very high temps . My laptop GT 520MX doesnt throttle even at 99C . But i m not that stupid and have underclocked it .
Answer is simple: Because it's free, saves electricity, puts less stress on ram/mobo/psu and ram is often next to cpu and heatsink which tend to get hot, so we really should reduce temps of ram memory considering that other components like GPU for example can transfer some heat to RAM modules. Also, ddr4 for example has default voltage of 1.2v, ddr1 has default 2.8v, it is obvious that Voltage is enemy to components and as technology develops, manufacturers tend to reduce voltage, heat and some ram modules even have their own heatsinks. Afterall, why wouldn't I do something to make my PC more stable and reduce temps drastically? Even VRM and Capacitors will last longer with all components undervolted (that includes capacitors in Power Supply) ... And btw, if you google about undervolting components, you will find out that it is VERY COMMON to UNDERVOLT :) There also exists undervolt overclocking where people manage to reduce voltage and increase frequency of CPU for example or GPU or RAM...
Early DDR3 was standard at 1.5V, but many later ones would run at 1.35V, like the DDR3 I used to have, which was 1,600 MHz. So a set doing 1,333 MHz at 1.2V doesn't surprise me too much, especially if you're using looser timings (can't remember what was considered good for DDR3 at a given speed but my 1,600 MHz RAM was 9-9-9 I think, so if your timings are similar or higher at 1,333 MHz, I'd say that's on the loose side and is helping them achieve it).

Same for the CPU, I forget what my 2500K was using. I want to say it was upper 1.1x or lower 1.2x range for 4 GHz but I'm not certain (I'm fairly certain it was mid 1.2x range at most, I know it wasn't near 1.3xV at all). If you're at stock, that voltage sounds like it might be about right.

The change from DDR2 to DDR3 probably wasn't the part what changed things most for you, but instead was what I imagine was a CPU change along with it?

I had a Core 2 which on a DDR3 platform (the Core 2 was originally DDR2 and was literally in the very last platforms given access to the then-new DDR3, and it often made little to no difference for it). Even on that PC, things are way, way, way slower than the 2500K system I moved to after, despite it technically being only a couple/few generations later (the entire first generation Core i series sort of was two if my memory isn't failing me, as it had a refresh within it, but maybe I'm wrong). The gap between those two platforms, in some ways, was bigger than the gap between that same 2500K and my current 3700X. Unless you were somehow using that 2600K on DDR2, I imagine it's more the rest of the platform change rather than going from DDR2 to DDR3 that uplifted your performance so much.
Last edited by Illusion of Progress; Nov 26, 2022 @ 1:42pm
UberFiend Nov 26, 2022 @ 1:41pm 
You're saving 1/4 the power of a watch battery, congratulations.
Dr. Danger Nov 26, 2022 @ 2:22pm 
Originally posted by UberFiend:
You're saving 1/4 the power of a watch battery, congratulations.
Are you aware that GPU core is usually around 1.2v for example and that if you manage to undervolt it to 1.050 you decrease temperature ALOT, something like from 75 to 65c and therefore prolong lifespan of that GPU. Manufacturers sometimes increase voltages on purpose because they don't want you to own a CPU or GPU that will be fast and still working after 10 years...
Last edited by Dr. Danger; Nov 26, 2022 @ 2:23pm
Dr. Danger Nov 26, 2022 @ 2:24pm 
Originally posted by UberFiend:
You're saving 1/4 the power of a watch battery, congratulations.
Now when I think about it... Watch battery uses much higher voltage than CPU or GPU or RAM and yet it has such a big influence on temps... Interesting.
76561198343548661 Nov 26, 2022 @ 2:28pm 
if you want to save power and reduce in case heat , get GTX 750ti . it s 6% faster , 2GB VRAM and 60W i.e. 128W less ! Issue solved .
Last edited by smallcat; Nov 26, 2022 @ 2:28pm
Dr. Danger Nov 26, 2022 @ 2:30pm 
Originally posted by smallcat:
if you want to save power and reduce in case heat , get GTX 750ti . it s 6% faster , 2GB VRAM and 60W i.e. 128W less ! Issue solved .
I am pretty much aware of that, but gtx 950 and 960 also have low wattage, only 1 6-pin connector compared to 2 6-pin that hd 5870 requires... gtx 950 is 95w so I think gtx 960 is 120w and my gpu is 180-200w...
76561198343548661 Nov 26, 2022 @ 2:35pm 
Jump to GTX 1050 ti , saving 113W being only 75W , nearly 2x faster , 4GB VRAM , so you re secured
Last edited by smallcat; Nov 26, 2022 @ 2:35pm
Originally posted by Dr. Danger:
Manufacturers sometimes increase voltages on purpose because they don't want you to own a CPU or GPU that will be fast and still working after 10 years...
You're right on the rest, but not this.

Offering a stable product is often the reason they don't run it on the edge of stability. An end user finding something is "stable" in whatever they decide to use to test it at doesn't mean it will actually remain stable in everything for the life of the part. As a result, the default voltage range would tend to skew towards the higher than "necessary" side.

What you said would make sense if most CPUs fail before getting replaced, but far more often they get replaced first. Failed CPUs are rather rare, actually, and I'd imagine the majority of those that do fail (beyond the ones that fail early due to more serious production defects) are likely those that are in the possession of owners who tinker with things like clock speed or voltage (usually pushing it up rather down, but still). Stock CPUs just don't fail often; they're typically replaced for, incidentally, not being fast enough/supported far before outright failure happens.
Last edited by Illusion of Progress; Nov 26, 2022 @ 2:38pm
UberFiend Nov 26, 2022 @ 2:39pm 
Originally posted by Dr. Danger:
Are you aware
I'm also aware of your previous threads & that you're aptly named.
Dr. Danger Nov 26, 2022 @ 2:41pm 
Originally posted by Illusion of Progress:
Originally posted by Dr. Danger:
Manufacturers sometimes increase voltages on purpose because they don't want you to own a CPU or GPU that will be fast and still working after 10 years...
You're right on the rest, but not this.

Offering a stable product is often the reason they don't run it on the edge of stability. An end user finding something is "stable" in whatever they decide to use to test it at doesn't mean it will actually remain stable in everything for the life of the part. As a result, the default voltage range would tend to skew towards the higher than "necessary" side.

What you said would make sense if most CPUs fail before getting replaced, but far more often they get replaced first. Failed CPUs are rather rare, actually, and I'd imagine the majority of those that do fail (beyond the ones that fail early due to more serious production defects) are likely those that are in the possession of owners who tinker with things like clock speed or voltage (usually pushing it up rather down, but still). Stock CPUs just don't fail often; they're typically replaced for, incidentally, not being fast enough/supported far before outright failure happens.
Yes, thanks for reminding me, totally forgot about that. Components degrade overtime and that means more voltage needed for that component to work. For example my undervolt might become unstable in a few years xD
Dr. Danger Nov 26, 2022 @ 2:42pm 
Originally posted by UberFiend:
Originally posted by Dr. Danger:
Are you aware
I'm also aware of your previous threads & that you're aptly named.
What do you mean by this? I mean, yes, I had some topics where I complained about my previous DDR2 configuration, maybe even my WiFi, but I did what everyone said - Upgraded.
_I_ Nov 26, 2022 @ 3:30pm 
how are the timings?
1333 cl11 is not very fast and should be able to run at 1.3v or lower
if its at 1333 cl7 and 1.4v that would be impressive
Dr. Danger Nov 26, 2022 @ 3:47pm 
Originally posted by _I_:
how are the timings?
1333 cl11 is not very fast and should be able to run at 1.3v or lower
if its at 1333 cl7 and 1.4v that would be impressive
it's 1333 cl9 by default, I was thinking about improving timings to cl7...
< 1 2 3 4 5 6 ... 10 >
Showing 1-15 of 137 comments
Per page: 1530 50

Date Posted: Nov 26, 2022 @ 11:48am
Posts: 137