How many DDR4 sticks should I go with?

[sc302 Veteran]

It really depends on the thermal capacity of the radiator.  The better it is at dissipating heat the quieter the fans, the less speed they need to cool the water. 

 

I had a very good setup with koolance. But the setup wasn't cheap.   Fans were always so low that the only fan you heard was the power supply fan.   The pump operation was also very quiet.  The radiator was huge in comparison to the cheap $100 kits.  In general, koolance is a 500+ investment...but you get what you pay for in quiet/near silent operation.  It just depends on how bad you want it. 

[Jim K Global Moderator]

32 minutes ago, Odom said:

Regarding your comment on the fans for the radiator, the case will already have two fans at the front sucking air in, plus another one at the back pushing it out. If the two radiator fans are also blowing air inwards, won't that be a bit too much?

Would it not make more sense having them blowing the air upwards (it would be mounted at the top of the BeQuiet Dark Base 900 case), so that the hot air will be expunged? This way there will be two fans blowing cool air inwards over the motherboard, and the other  ubee pushing warm air from the inside to the outside.

Early, haven't had all my caffeine injections yet.  My recommendation is if you go with the long (dual) radiator ... stick it up front if feasible (you'd probably lose some drive bays) ... blowing into the case.  If you go with a single radiator ... stick it on the back exhaust blowing out.  I only have a single radiator (H80) ... have it mounted on the back ... and my CPU stays nice and cool.

 

Also...this is a gaming PC for the living room ... right?  Not just a PC for watching movies (i.e. HTPC)?

 

Per their website regarding radiator support for the 900...

Front (mm)

120 / 140 / 240 / 280 / 360 / 420

Top (mm)

120 / 140 / 180 / 240 / 280 / 360 / 420

Bottom (mm)

120 / 140 / 280

Rear (mm)

120 / 140

[Odom Member]

Yes, mainly a gaming PC. It will be at one wall in the living room and my brother has an HDMI cable connecting to a 4K TV on the opposing wall, so he will be watching movies as well. But because the PC will be in the living room, essentially behind the couches, I would like to try and get it quiet.

 

Thanks for the hints! I'll try and see how to best fit the components into the case and whether or not I'll be going with a single or dual radiator.

[n_K]

On 30/06/2016 at 8:56 PM, Astra.Xtreme said:

Also, not sure how hot RAM sticks get these days, but less sticks = less heat.

No that's nonsense. The heat is due to two reasons:

1. The switching effect of the components (this cannot be stopped, though it can be reduced with lower voltages)
2. The inefficiencies of the components (this can be greatly reduced by using better quality materials).

More slots uses does not mean more heat, inefficient RAM means more heat.

[Jim K Global Moderator]

Just now, Odom said:

Yes, mainly a gaming PC. It will be at one wall in the living room and my brother has an HDMI cable connecting to a 4K TV on the opposing wall, so he will be watching movies as well. But because the PC will be in the living room, essentially behind the couches, I would like to try and get it quiet.

 

Thanks for the hints! I'll try and see how to best fit the components into the case and whether or not I'll be going with a single or dual radiator.

aaah ok...and if I had read one of your early posts I would have seen that.   (like I said...it's early)

 

What GPU are you going to stick in it?  1070/1080??   

[DevTech]

35 minutes ago, Odom said:

 

Regarding your comment on the fans for the radiator, the case will already have two fans at the front sucking air in, plus another one at the back pushing it out. If the two radiator fans are also blowing air inwards, won't that be a bit too much?

Would it not make more sense having them blowing the air upwards (it would be mounted at the top of the BeQuiet Dark Base 900 case), so that the hot air will be expunged? This way there will be two fans blowing cool air inwards over the motherboard, and the other three pushing warm air from the inside to the outside.

 

It is a good idea to forget about visualizing airflow as if air obeys the logic of a human brain. Every test I have seen where the actual heat map is measured with a Infrared sensor shows that out intuition is almost always wrong. One important reason is that all of the computer cases are leaky buckets (think of the rubber tube in a bicycle tire full of small holes) and essentially have close to zero suction. The fan is just as likely to suck air from a nearby joint in the case as hot air inside. The only certainty if you are planning cooling without a IR Sensor Gun is that if you blow air directly onto something that is likely to work the way you think it should. Because a case is so leaky, the air you blow in always leaks out somewhere, there is never any mythical "positive case pressure"

 

So the idea is to have air flow over any component that needs cooling with the minimum number of quiet slow moving fans. Fans that suck air out are just not very efficient and really just add to the overall noise level. Remove those fans and it leaves extra space for the air to exit.

 

Since the CPU is the main source of heat, it is more important to keep that cool than worry about the general case temperature. In some cases the GPU becomes the priority. So basic physics is that if you are blowing air through something, you can't get cooler than the temperature of the air you are blowing. Therefore in the case of the CPU cooling it almost NEVER makes sense to use internal case air as the source of CPU cooling. So unless you have a component that generates more heat than the CPU, the CPU should get priority for cool air and then adapt all the cooling strategy around that. From experience, besides the CPU radiator, all you need is 1 fan blowing air onto the motherboard  to cool the CPU power supply and 1 fan blowing air across the hard drives and very careful analysis to determine if the GPU can benefit from getting outside cool air from somewhere. Most GFX cards make that very difficult.

 

[Odom Member]

My brother currently has my older GTX680 (maybe Ti? I don't rember anymore).

I currently have the GTX970 and am still thinking if I get the GTX1080 or wait for the GTX1080Ti. Once I get a new one for myself he will get my GTX970. I think that should be good enough for him to play games maxed out at 1920x1200. I can play most games almost at max settings on 3440x1440, so I think he should be fine at the lower setting.

[DevTech]

2 minutes ago, Odom said:

My brother currently has my older GTX680 (maybe Ti? I don't rember anymore).

I currently have the GTX970 and am still thinking if I get the GTX1080 or wait for the GTX1080Ti. Once I get a new one for myself he will get my GTX970. I think that should be good enough for him to play games maxed out at 1920x1200. I can play most games almost at max settings on 3440x1440, so I think he should be fine at the lower setting.

One thing I have noticed with QHD type screens is that FSAA becomes marginal in improving appearance and turning it off completely really boosts the GPU's processing ability for everything else.

 

[Astra.Xtreme]

On 7/2/2016 at 7:58 AM, n_K said:

No that's nonsense. The heat is due to two reasons:

1. The switching effect of the components (this cannot be stopped, though it can be reduced with lower voltages)
2. The inefficiencies of the components (this can be greatly reduced by using better quality materials).

More slots uses does not mean more heat, inefficient RAM means more heat.

It's not nonsense.  It's common sense actually.

 

Each RAM stick is typically powered by something around 1.2-1.5V, right?  The more sticks you add, the higher the capacitive load it will be.  Basic electrical theory states that if voltage stays constant, as the load increases, more current is required to keep the voltage constant.  More current draw = more heat.  That's straight up physics.  Ever touch your RAM sticks after playing a memory taxing game?   They get decently hot.  It's sort of why most RAM sticks have big heatsinks attached to them.  I'm not saying that 4 sticks vs 2 sticks will be a massive heat difference, but it's definitely measurable. 

 

What you said about inefficiency is actually nonsense.  There is no such thing as a more "efficient" RAM stick.  There actually are only a few suppliers of SDRAM chips in the world, so all brands of RAM are almost the same in terms of the components that they're made of.  Plus all RAM manufactures are RoHs compliant, so they will all use the same materials to piece the components together.

[n_K]

8 hours ago, Astra.Xtreme said:

It's not nonsense.  It's common sense actually.

 

Each RAM stick is typically powered by something around 1.2-1.5V, right?  The more sticks you add, the higher the capacitive load it will be.  Basic electrical theory states that if voltage stays constant, as the load increases, more current is required to keep the voltage constant.  More current draw = more heat.  That's straight up physics.  Ever touch your RAM sticks after playing a memory taxing game?   They get decently hot.  It's sort of why most RAM sticks have big heatsinks attached to them.  I'm not saying that 4 sticks vs 2 sticks will be a massive heat difference, but it's definitely measurable. 

 

What you said about inefficiency is actually nonsense.  There is no such thing as a more "efficient" RAM stick.  There actually are only a few suppliers of SDRAM chips in the world, so all brands of RAM are almost the same in terms of the components that they're made of.  Plus all RAM manufactures are RoHs compliant, so they will all use the same materials to piece the components together.

ROHS has nothing to do with efficiency. Different RAM chips are produced different and no it's complete nonsense to say they're all the same: different fab size (makes a HUGE difference to efficiency), different clocking speed, different refresh rates, different latencies, different internal structure, etc.

If you're so sure that memory chip efficiency makes no difference (you're completely wrong) let's compare a DDR4 stick https://www.micron.com/~/media/documents/products/data-sheet/modules/parity_rdimm/asf9c512x72pz.pdf with some old 1KB SRAM http://www.nteinc.com/specs/65000to65999/pdf/nte65101.pdf

4GB DDR4 stick: Active standby current IDD3N 603-567 mA

old 1KB SRAM: 9.0-22 mA

1KB*4194304 = 4GB -> 9.0-22*4194304 = 37-92 MA

That's MA not mA, MA is 1x10^6, mA is 1x10^-3 But right there's no such things as memory efficiency!111one

 

The current in DC is determined by the resistance, not the capacitance, this is basic physics.

[Mindovermaster Global Moderator]

5 hours ago, n_K said:

ROHS has nothing to do with efficiency. Different RAM chips are produced different and no it's complete nonsense to say they're all the same: different fab size (makes a HUGE difference to efficiency), different clocking speed, different refresh rates, different latencies, different internal structure, etc.

If you're so sure that memory chip efficiency makes no difference (you're completely wrong) let's compare a DDR4 stick https://www.micron.com/~/media/documents/products/data-sheet/modules/parity_rdimm/asf9c512x72pz.pdf with some old 1KB SRAM http://www.nteinc.com/specs/65000to65999/pdf/nte65101.pdf

4GB DDR4 stick: Active standby current IDD3N 603-567 mA

old 1KB SRAM: 9.0-22 mA

1KB*4194304 = 4GB -> 9.0-22*4194304 = 37-92 MA

That's MA not mA, MA is 1x10^6, mA is 1x10^-3 But right there's no such things as memory efficiency!111one

 

The current in DC is determined by the resistance, not the capacitance, this is basic physics.

There a reason why you are comparing 2 entirely different types of RAM? Yes, it was different 20 years ago....

[n_K]

3 minutes ago, Mindovermaster said:

There a reason why you are comparing 2 entirely different types of RAM? Yes, it was different 20 years ago....

http://www.samsung.com/semiconductor/global/file/product/DS_64GB_TSV_DDR4_4Gb_D_die_RegisteredDIMM_Rev13.pdf Page 26, go compare your hearts content out with the micron RAM and notice how it's completely different!

[Mindovermaster Global Moderator]

1 hour ago, n_K said:

http://www.samsung.com/semiconductor/global/file/product/DS_64GB_TSV_DDR4_4Gb_D_die_RegisteredDIMM_Rev13.pdf Page 26, go compare your hearts content out with the micron RAM and notice how it's completely different!

Really, REALLY, who gives a ******? Does it really matter in real life scenarios? No. Everyone has their favorite RAM company. Prices fluctuate every day.

[Jim K Global Moderator]

Four sticks will generate more heat ... because well two sticks get generally warm/hot ... so you're basically doubling that.

 

Does it matter and/or affect performance/longevity?  NOPE

 

I would still populate two slots with higher capacity modules ... vs. filling 4 slots with lower capacity.  Two high capacity modules usually run cheaper than 4 lower capacity ones.  Plus it gives you room to expand in the future.  Heat doesn't at all factor into the reasoning.  So I would do 2x8GB before doing 4x4GB

[Daedroth]

I've always preferred fewer higher capacity sticks over more lower capacity sticks. Reason being is that it'll take more for one stick to become full and another being required. Whether that has any basis in actual performance performance, I don't know, but it's how I justify it!

[Mindovermaster Global Moderator]

I always do 2 sticks. Because my aftermarket CPU heatsink normally takes up the closest slot near the CPU...

 

I remember my 775 computer had a problem with 4 sticks... Won't run 'em...

[Astra.Xtreme]

7 hours ago, n_K said:

ROHS has nothing to do with efficiency. Different RAM chips are produced different and no it's complete nonsense to say they're all the same: different fab size (makes a HUGE difference to efficiency), different clocking speed, different refresh rates, different latencies, different internal structure, etc.

If you're so sure that memory chip efficiency makes no difference (you're completely wrong) let's compare a DDR4 stick https://www.micron.com/~/media/documents/products/data-sheet/modules/parity_rdimm/asf9c512x72pz.pdf with some old 1KB SRAM http://www.nteinc.com/specs/65000to65999/pdf/nte65101.pdf

4GB DDR4 stick: Active standby current IDD3N 603-567 mA

old 1KB SRAM: 9.0-22 mA

1KB*4194304 = 4GB -> 9.0-22*4194304 = 37-92 MA

That's MA not mA, MA is 1x10^6, mA is 1x10^-3 But right there's no such things as memory efficiency!111one

 

The current in DC is determined by the resistance, not the capacitance, this is basic physics.

Why did you even bother rambling about obsolete RAM vs modern RAM?  Do you, for some reason, think that a modern DDR3 or DDR4 RAM stick is going to have old SDRAM chips on it?  You've deflected the discussion to something completely irrelevant...

 

Here's a list of things for you to research,and maybe you'll understand this topic a little better:

1.  Energy consumption of a capacitive load.  I don't think you realize that electrons in AC are the same as electrons in DC.  Capacitors charge and discharge energy. More capacitors = more additive energy being discharged.  Power is energy dissipated over time.  Guess what power is?

 

2. Look up a list of SDRAM suppliers.  Then look up what SDRAM chips that are used in the varieties of PC RAM sticks.  Maybe you'll realize something when you see the short list.

 

3. Learn what RoHS is.  The point I made clearing went over your head.

 

If you honestly don't believe any of this, go buy a Kill-a-Watt and do the measurements yourself.  Run a full blown benchmark test with 2x4GB vs 4x2GB and you'll see a several Watt difference. It's not magic or anything?  Again, I'm not saying it's anything substantial.  We're probably talking $5 or less per year in energy difference, but it's absolutely not nonsense.  The facts prove that.

[n_K]

1 hour ago, Astra.Xtreme said:

Why did you even bother rambling about obsolete RAM vs modern RAM?  Do you, for some reason, think that a modern DDR3 or DDR4 RAM stick is going to have old SDRAM chips on it?  You've deflected the discussion to something completely irrelevant...

 

Here's a list of things for you to research,and maybe you'll understand this topic a little better:

1.  Energy consumption of a capacitive load.  I don't think you realize that electrons in AC are the same as electrons in DC.  Capacitors charge and discharge energy. More capacitors = more additive energy being discharged.  Power is energy dissipated over time.  Guess what power is?

 

2. Look up a list of SDRAM suppliers.  Then look up what SDRAM chips that are used in the varieties of PC RAM sticks.  Maybe you'll realize something when you see the short list.

 

3. Learn what RoHS is.  The point I made clearing went over your head.

 

If you honestly don't believe any of this, go buy a Kill-a-Watt and do the measurements yourself.  Run a full blown benchmark test with 2x4GB vs 4x2GB and you'll see a several Watt difference. It's not magic or anything?  Again, I'm not saying it's anything substantial.  We're probably talking $5 or less per year in energy difference, but it's absolutely not nonsense.  The facts prove that.

1. A capacitor acts completely different under AC than it does DC. DC is stores a charge acting as a temporary energy store, AC is passed right through a capacitor. The difference in capacitance from 2vs4 (where the 2 sticks is double the density of the 4 sticks) sticks of RAM would be negligible and down to the traces.

2. It doesn't matter if there's 1 supplier of chips of 100, every single damn different product is different. If all RAM chips each company produced were the same then they'd only produce ONE RAM CHIP because it wouldn't make any sense to batch produce 5 chips with the exact same characteristics would it?

3. Evidently you don't have a clue about it. It has nothing - nothing - to do with the efficiency of RAM but if you think you can post a scientific paper on it I'd be more than happy to read it. And no, tin whiskers is not even relevant.

A power meter is accurate to 5-10% which means readings within a few w are worthless (again this is literally entry level physics stuff). And let's assume for a moment you are right and you're saying using 4 sticks consume 5w more, DDR4 is about 1.2v so 1.2A = 5 -> 5/1.2=A -> A = 4... You are claiming using 2 extra sticks uses 4 AMPS more power? Read the ######## datasheets I posted you complete plonker, the current consumption of RAM whilst active is measured in mA, so you're completely off your rocker.

And let me make it clear to you in very simple terms so even simple simon can get it: if a stick of 1GB DRAM has let's say ~1m capacitors for memory retention, then a 2GB DRAM DIMM would have double that, ~2m... Oh gee, the difference in number of capacitors of 2x1GB being 2*1m vs 1x2GB being 1*2m... Oh would you look at that, 2m capacitors regardless of which one you pick! Wow! Gosh! Gee! The actual important part of the capacitors in RAM is how often they need to be refreshed (aka how fast they leak charge), the less refreshes they need the less power needs to be used recharging them up.

[Astra.Xtreme]

11 minutes ago, n_K said:

1. A capacitor acts completely different under AC than it does DC. DC is stores a charge acting as a temporary energy store, AC is passed right through a capacitor. The difference in capacitance from 2vs4 (where the 2 sticks is double the density of the 4 sticks) sticks of RAM would be negligible and down to the traces.

2. It doesn't matter if there's 1 supplier of chips of 100, every single damn different product is different. If all RAM chips each company produced were the same then they'd only produce ONE RAM CHIP because it wouldn't make any sense to batch produce 5 chips with the exact same characteristics would it?

3. Evidently you don't have a clue about it. It has nothing - nothing - to do with the efficiency of RAM but if you think you can post a scientific paper on it I'd be more than happy to read it. And no, tin whiskers is not even relevant.

A power meter is accurate to 5-10% which means readings within a few w are worthless (again this is literally entry level physics stuff). And let's assume for a moment you are right and you're saying using 4 sticks consume 5w more, DDR4 is about 1.2v so 1.2A = 5 -> 5/1.2=A -> A = 4... You are claiming using 2 extra sticks uses 4 AMPS more power? Read the ######## datasheets I posted you complete plonker, the current consumption of RAM whilst active is measured in mA, so you're completely off your rocker.

And let me make it clear to you in very simple terms so even simple simon can get it: if a stick of 1GB DRAM has let's say ~1m capacitors for memory retention, then a 2GB DRAM DIMM would have double that, ~2m... Oh gee, the difference in number of capacitors of 2x1GB being 2*1m vs 1x2GB being 1*2m... Oh would you look at that, 2m capacitors regardless of which one you pick! Wow! Gosh! Gee! The actual important part of the capacitors in RAM is how often they need to be refreshed (aka how fast they leak charge), the less refreshes they need the less power needs to be used recharging them up.

1.  Please explain how power being dissipated is magically different...  The functionality of a cap in AC and DC is completely irrelevant.  Power consumption is power consumption.  Not sure why you can't comprehend such a simple concept...

2.  This goes back to your "efficiency" claims nonsense.  A Gskill DDR3 RAM stick is going to be almost identical to a Kingston DDR3 RAM stick.  That's due to only handful of suppliers of SDRAM.  There is no magical efficiency advantage to one over the other.  There is no proof of any of that.  Again, not a hard concept, and stop trying to divert from the original point...

3.   Again, you completely neglected the point...  That tactic is getting stale...  There are plenty of very accurate power meters out there.  I have one sitting 3 feet from me. Try comprehending the point for once...  You REALLY need to rethink your idea of how you're calculating the current draw.  Where do you think the RAM sticks are pulling the current from?  Oh yeah, the motherboard...  And what voltage is the motherboard running on? Do you math much? 

 

It's not worth beating the dead horse anymore since you have absolutely no clue about any of this...  It's sad that I even had to waste this much time explaining these things to you.  Next time open a book or use some common sense.  This all comes back to your failed attempt at calling me out: "More slots uses does not mean more heat, inefficient RAM means more heat."  Completely and absolutely wrong...  Nice diversion attempts, but you're wrong.  Get over it...  I'm sure you'll have some witty and irrelevant statement to make, but don't bother.   I'm moving on to something intelligent...