• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Can we actually get some definitive answers on B Die temperature?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

gameinn

Registered
Joined
Jul 1, 2020
I don't see this mentioned in much reviews or anything. Either the reviewer said the overclock worked or not and I know there are many people here who tested much memory kits.

Most information you find on this subject is random reddit threads where the safe temperature values vary from people saying below 40c is fine all the way up to 60... that is quite a delta.

My other issue comes from the steadily increasing size and power of GPU and the temperatures they bring. For example if you had a GTX 1660 Ti and then somehow went to a 3070 or 3080 your RAM is probably having a far worse time due to the way the cooler on those operates.

Basically we need an indepth 2021 guide to B Die temperature based on using current gen GPUs. Bad bin vs good bin, does the bin quality effect voltage tolerance? If so, why does it? Are other brands better at binning for voltage tolerance? Can the thermal sensor in say G. Skill kits be accurate enough or acual thermal couples needed for measurement?
 
Last edited:
search this site for "Woomacks memory test list" has a ton of info on most memory and chips on them. He is very knowledgeable with all of it.
 
Here is your answer: it depends. If you want to know what is safe, 60 is fine. However once you start pushing the OC, B-die can become unstable due to temperature at lower temps.

I'm not sure that GPU power draw is really steadily increasing. Sure the 3090 is a beast, but I have a 2 year old card pulling 360W at times. You can't really compare the 1660Ti to a 3070 or 3080, those cards are in a different class. The 2080 might have drawn a bit less than the 3080 depending on BIOS and overclocking, but the 3070 and 2070 are nearly identical in power draw. In other words you're getting the same performance as the 2080Ti for 125W less power in the 2070.

In terms of the cooler design, it's not a big deal. Since the exhaust air is still going to be lower temperature than the goal temp for the memory, increased airflow can actually still help the memory. Also this is far from the first card to have a flow through cooler. nVidia made such a big deal about it in the reference design that board partners feel obligated to do it. At the end of the day, overall good airflow in the case is going to make a bigger difference than GPU cooler design.
 
I don't consider temperature of ram near as much as temperature of other pieces because I don't think it is important. In my experience I've had memory blazing hot to the touch but ran just fine and some cool as ice and completely unstable. Your CPU gets too hot it will stop functioning or shut down, similar to the GPU which will crash or restart. Memory will either run the speed with the volts or it won't. I can't think of a time where I had memory heat be a problem so I wouldn't worry about temps if I were you. If you don't like the heat it is adding stick a fan blowing over them. It will cool the whole area which in turn will cool your GPU if that is what you are worried about.

Not to mention decent memory is what $120US. If it cooks then I'll buy more.
 
No one cares about temps because if you run it in a reasonable environment then DDR4 never reaches ~60°C under a long 100% load, without additional airflow. Max safe temps for consumer-grade DDR4 are 85°C. Dual rank modules heat up about 5°C more so instead of typical ~45°C under load, you get ~50°C.
I was many times mentioning temps in my reviews but most of my reviews are not including Samsung B IC. There are so many other and more interesting IC that I try to provide a wide range of various memory kits. I don't like to repeat the same stuff for longer. Not to mention that Samsung B is on the market for a couple of years and this topic was covered multiple times on the OCF and a thousand times around the web.

Basically we need an indepth 2021 guide to B Die temperature based on using current gen GPUs. Bad bin vs good bin, does the bin quality effect voltage tolerance? If so, why does it? Are other brands better at binning for voltage tolerance? Can the thermal sensor in say G. Skill kits be accurate enough or acual thermal couples needed for measurement?

Really, no one besides you needs that. Also, no one will provide results based on good and bad bins. To get that info, you have to run IC at really tight timings and high voltages and have specific top overclocking motherboards. 99%+ users don't care and won't see the difference. Most B-die will run the same at popular and recommended settings, regardless if it's a good or bad bin. Bad bin won't run at 4000 CL12-12-12 or 4800+ CL18. Really who cares about that? ...
Soon there will be DDR5 on the market. Making another B-die guide right now is a total waste of time.
 
Last edited:
Sounds like you could be moving more air. I was running 1.5v on my sticks for a good chunk of the day yesterday, and didn't see anything over 30 on them with just case flow.
 
Typically it's like 30-35°C idle or with low load (games are usually still low load) at 1.35-1.45V. When you run it in something really demanding, doesn't have to use the whole RAM capacity, then it will go up to ~50°C. That's for single rank modules with average airflow or at least not many other things heating air up. When you run dual-rank modules in the same tests, then in idle, temps will be similar but under full load will be about 5°C higher.

Recently I was testing 2x32GB/dual-rank Microns at 1.45V in a really small ITX case, with limited airflow and no exhaust fan. I was able to reach ~70°C. That was the absolute max after 3h+ high CPU+RAM+GPU load where the CPU was constantly at ~90°C and gfx card was only GF1030 but with passive cooling so the air was barely moving inside.

The heat affects overclocking but it doesn't really matter at voltages up to ~1.6V. In competitive overclocking, you can use additional fans or LN2 or whatever you like for these additional couple MHz or one CL less. In typical daily usage, even overclocked but at reasonable voltages, it doesn't really matter.
 
Not to mention that Samsung B is on the market for a couple of years and this topic was covered multiple times on the OCF and a thousand times around the web.

Just because a topic was covered by random users on the internet doesn't mean it was actually covered. That's like me going into random reddit threads and stating 70c is fine on b die. Eventually this becomes the truth. I'm sure most people would agree this has happened for some hardware topics.

As an example let's take the talks over 2 vs 4 stick of ram that still goes on to this day due to this video:

One YouTube video literally made 90% of the pc community state a mantra of "4 sticks is better than 2". We know this to be false but if 1 person says it the sheep will follow. I am stating this as to why posts about temperature might also fall into this flow of misinformation. It's not exactly something easy to articulate so my vetting of info is much higher than say a reddit user telling me 2+2 = 4.

To get that info, you have to run IC at really tight timings and high voltages and have specific top overclocking motherboards. 99%+ users don't care and won't see the difference

Well I mean we are you know on a forum called overclockers, lol. Plus based on that logic why do you even overlcock RAM then? 99%+ users are fine with RAM at 3,200 16-18-18-38 so why give info that 99%+ users won't see any difference from? It can work both ways.

Soon there will be DDR5 on the market. Making another B-die guide right now is a total waste of time.

2022 is soon? I know late 2021 MIGHT happen for DDR5 but this was probably before covid. A guide that takes maybe a week to do the research and write that is going to be pertinent for close to 2 years seems insane value to me. Again let's be honest most users won't be using DDR5 anyway for atleast 3+ years. I have no inside knowledge but $300 for 16GB DDR5 seems like a decent starting point... most users won't spend $50 on RAM. :D

Sounds like you could be moving more air. I was running 1.5v on my sticks for a good chunk of the day yesterday, and didn't see anything over 30 on them with just case flow.

Oh no I have literally 0 concerns about temperatures for my PC or anything like that. I just wanted to learn something else for today to improve my own knowledge in the vast pool of overclocking. It seems like my post brought up alot of hostile replies (from my interpreatation) "Really, no one besides you needs that."... lol. Anyway just thought I'd post this to let people know where I am coming from but I won't be back after posting this. I'm tired of having to solve captchas every few days to even access this website. Peace.
 
You expect someone to make a guide for memory IC which is on the market for a couple of years. Just because it's a popular IC for new Ryzens, it doesn't mean it's required to cover the same topic again. It's also a very narrow topic which can be answered in 2 sentences so I see no point in writing guides (and I already answered to that in my previous posts).
Personally I skip Samsung B because it's boring and I've written at least 20 reviews about this IC. I don't know if you realize that no one is getting paid for reviews or guides here besides some review samples from time to time that they can keep (not even all). For me it's like I test RAM for fun. I spend my money on that too and sometimes I buy memory to show something interesting or I ask vendors to send me specific kits for tests (as I really don't earn much so it's hard for me to cover hardware for tests). Tests based on binning would require multiple memory kits, like really many to make a good comparison. It's simply not possible and for sure not worth to do considering already known results. I won't even mention how many hours would take really honest test with full comparison etc.

There were full Samsung B overclocking guides on extreme OC websties. I guess there is still something in the archives of hwbot forums. However, these guides were made for extreme overclockers who run RAM at 1.9V+. In some of my reviews are results up to 2V+ on Samsung B but if the memory kit is not worth it, then I simply don't push it.
I realize that this website is called Overclockers. However, I don't know if you noticed but overclocking is dying everywhere. From a large group of really extreme guys, we have only few left. Those who left know what they are doing and don't need guides. All others don't care or enjoy reading about these things but will never take part in extreme overclocking. So reviews or guides are more for enthusiasts and gamers but more detailed or more extreme OC is only when it's worth to do that or reviewer actually enjoy the product enough to spend additional time.

In many reviews I said to ask questions or suggest what do you want to see in reviews. I mean additional tests, some specific comparisons etc. I was asking that for years and not even one person suggested anything. All those who have questions, ask them in forum reviews or send me questions directly. For some reason, many forum users send me pms or emails with questions instead starting a new thread or post in the specific memory thread. Many make forum accounts only to send me questions via online form. One of reasons is probably like your issue right now - you can't find any good info in the web and posting on the forums will end only on wasting your time and reading answers of clueless people who won't even test anything before answering (I mean mostly other forums as I know that most people come here when they had bad experiences in other places).

I'm sorry if you took my replies as hostile. I'm not a native English speaker and I could use a wrong form. Anyway, the thread name and expectactions in the first post suggest that you want that someone write a guide what was in the form of demand rather than request.
I can agree with the fact that the web is flooded with random info which is often false. People repost other people's thoughts and never test anything on their own. Many create theories which aren't true or even close to reality. Later you can't find any info on even these popular, and for some reason, respected websites. More often, frontpage reviews are really poor quality and include nothing valuable. It's harder to find good info about products you want to buy. Youtube reviews are a joke and it's mostly flooded with "unpacking reviews" like anyone needs that.
 
I feel like some of this is the result of this one size fits all expectation for overclocking. Nobody wants to make incremental changes and test for stability any more. A guide can only go so far, at the end of the day you're going to to have to see what works for your hardware.
 
Where does my B-Die memory module become unstable due to temp?

What is the exact temp this "B-die" can handle?

****, who knows? One stick might get flaky at 43c and another at 61c.

That's why you read such wide numbers.

So 40-60c is both true and false at the same time.

Simple rule of thumb with overclocking.

Rule 1: when you increase frequency and voltage, you lower temperatures.

That's the goal. Want faster, get colder. Period. (Not all HW, but we can say 90% of it.)

Hope this helps!
 
If at first you aren't succinct, call ShrimpBrime!
 
Back