Question Time to upgrade to dual channel?

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
Hello everyone!

I just upgraded to an RTX 3070 and performance is great. However, there are several games where I do not get 99% GPU usage. I'm not sure whether it's a CPU bottleneck or a RAM bottleneck, but it seems to be worse in older games where single core performance is more important, but also, my 8700K should be doing better with them. So I am considering whether or not my single channel DDR4 2666 would benefit from another stick so that I can be dual channel. The internet is incredibly bipolar about this. YouTube videos will show 20% - 40% where there's a RAM bottleneck while some articles will have charts and graphs proving that dual channel has no benefit?

Can someone help me out? I know my setup is old, 2666 is slow by today's standards, and single channel may be more of an issue now.

Also, while I'm got you guys here, any chance it's a hard drive bottleneck? That mostly shows up as stuttering right? Because I ran a lot of the games from 2015-2019 from an external HDD with 150 - 170 MBps because 10TB externals are so cheap.
 
Hello everyone!

I just upgraded to an RTX 3070 and performance is great. However, there are several games where I do not get 99% GPU usage. I'm not sure whether it's a CPU bottleneck or a RAM bottleneck, but it seems to be worse in older games where single core performance is more important, but also, my 8700K should be doing better with them. So I am considering whether or not my single channel DDR4 2666 would benefit from another stick so that I can be dual channel. The internet is incredibly bipolar about this. YouTube videos will show 20% - 40% where there's a RAM bottleneck while some articles will have charts and graphs proving that dual channel has no benefit?

Can someone help me out? I know my setup is old, 2666 is slow by today's standards, and single channel may be more of an issue now.

Also, while I'm got you guys here, any chance it's a hard drive bottleneck? That mostly shows up as stuttering right? Because I ran a lot of the games from 2015-2019 from an external HDD with 150 - 170 MBps because 10TB externals are so cheap.
Yes, dual channel RAM is always better,up to 50% depending on some other factors but same question is how much total memory performance affects particular program/game.
External drives usually suffer from slow transfer speed an high system (mostly CPU) load while reading or writing. Even internal slow drives can do that.
 
Reactions: Roland Of Gilead

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
My build is an 8700, RTX 3070, 16GB Hyper X Fury DDR4 2666 (1 stick), my power supply is an EVGA SuperNOVA G6 750 Wat bought one month ago.

"Yes, dual channel RAM is always better,up to 50% depending on some other factors but same question is how much total memory performance affects particular program/game. "

Which is why I'm here, to figure out if it is needed for my gaming needs, considering everyone have conflicted answers. I don't want to upgrade RAM if the issue is unfixable since the games are just not threaded well.

"External drives usually suffer from slow transfer speed an high system (mostly CPU) load while reading or writing. Even internal slow drives can do that. "

Of course, but would that cause low GPU usage? It doesn't appear that the CPU overall is being taxed or that the HDD is too slow, since newer games run fine with 99% GPU usage and high CPU usage.
 

geofelt

Titan
First of all, you should not want your gpu to run at 99%
That would indicate that your gpu was limiting your gaming.
Your upgrade is obviously working.

There are several benefits from upgrading ram, and it depends on the problem you are trying to fix.

1) If you have insufficient ram for your workload, it will show up in task manager/resource monitor/memory tab/ hard faults.
Anything more than zero suggests you need more ram.

2) Dual channel operation is faster. But not as much as you may think.
It might be the equivalent of 2% faster cpu operation in games.

3) Faster ram improves cpu performance, but not so much on intel systems.
Perhaps similar to going to dual channel.

What is the make/model your motherboard?
What is the make/model of your current ram?
What is your cpu cooler?
I ask because even a conservative overclock of a 8700K can gain some 25% cpu performance.
You would want a decent cpu cooler to do so.

Lastly, simply adding a ram stick is not guaranteed to work.
Your odds of success are perhaps 90%
It is ok to try if you can reasonably return the ram.
It might be better to buy a 2 x 8/16gb matched kit.

Stuttering is caused by a temporary lack of a critical resource.
It usually shows up as a lack of cpu, but it could also be hard faults due to insufficient ram.

Hard drives are some 40x slower than a ssd in random operations, and 3-6x slower in sequential operations.

If you do not have a decent sized ssd for your C drive, that is what I would look to as a first upgrade.

Do you have an upgrade budget?
 
Hey there,

As others have pointed out, dual channel gives much more performance, more often than not. Others are incorrect in their suggestions.

Here's an example of the difference beteen single channel and dual channel configs in terms of gaming. The increase is a lot more than 2%. Not always, as each game or app responds differently to the increase in mem bandwidth. Not only higher fps but better mins/avg/1.0% lows. It feels better.Dual channel vs Single channel RAM | Intel 9th gen i5 9400F - YouTube

For exammple DDR4 2133 in single channel mode offers 17gb/s of mem bandwidth, versus dual channel at 34gb/s. Why would you not want the extra bandwitdh to maximize your system performance? There is no downplaying it, as if it's not needed. Any descerning gamer/system builder would do the same. The main reasoning is this: our CPU (8700k )has a maximum bandwitdh of 41.6gb/s. You are feeding it approx 20gb/s. That's the bottleneck! Your CPU is capable of driving the 3070. OC it a bit when you get this issue sorted.

Buy a 2 x 8gb matched kit of 3200mhz ram, and you can judge for yourself the increase you will experience.

I think in general terms, we are all saying the same thing to you :)
 
Last edited:

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
"First of all, you should not want your gpu to run at 99% "

i understand, but when I'm only getting 100-120 in old FPS games at 144 Hz refresh with 75% GPU usage, makes me wonder if there's something I need to do to fix it. When I max out graphics in a newer game, I'll probably get close to 100%, like with Control with Raytracing. But, then, I'll only get 80% with Quantum Break and 90-100 FPS ( to compare games by the same developer).

"What is the make/model your motherboard? "

An HP Omen prebuilt motherboard of some sort (got a super good deal in late 2019, had to, never would've gotten a prebuilt otherwise). and it is a non-K version so I cannot overclock. And I can get more RAM sticks from Crucial which are guaranteed to work with my model so that's not an issue.

I'm just trying to figure out why, say, I get, say, 70% GPU usage in some older games. I get that the graphics card is powerful for some of these games, and they are not threaded like more recent games, but I just don't understand why the GPU is low while still being under my refresh rate on games that even an 8700 ( I assume) should be able to get more FPS. I could be wrong, make its just the single core performance of my CPU, but I want to make sure.

"Stuttering is caused by a temporary lack of a critical resource. "

To be clear, I have zero stuttering issues. I am asking how a HDD bottleneck would present itself since I am also considering whether or not running from an external HDD is causing the bottleneck.

"If you do not have a decent sized ssd for your C drive, that is what I would look to as a first upgrade. "

My Windows runs on an SSD, my primary games run on an internal hard drive, superfluous games that I don't play as much are on an external since space is so cheap.

"Buy a 2 x 8gb matched kit of 3200mhz ram, and you can judge for yourself the increase you will experience. "

But by then I've already spent the money, and if it's just that I am playing old games then the issue isn't fixed. As I said, I get max GPU usage in newer games, but at lower FPS which wouldn't really test the CPU/RAM anyways.

"Here's an example of the difference beteen single channel and dual channel configs in terms of gaming. The increase is a lot more than 2%. Not always, as each game or app responds differently to the increase in mem bandwidth. Not only higher fps but better mins/avg/1.0% lows. It feels better.Dual channel vs Single channel RAM | Intel 9th gen i5 9400F - YouTube "

Yep, I've seen that exact video. The problem is in atleast a few of those games, such as AC Odyssey, I DO get max GPU usage.
 
Last edited:
Note, I've not mentioned GPU usage in my post once. This is almost a misleading metric.

Edit: Yes, with a really powerful GPU, you can expect low usage in some games. This can be from a combination of reasons. The game engine might be capped, the graphics in the game are very basic, or older and don't require as much horsepower. It can also be down to power plans running.
But for the most part, you do want your GPU at 99% or it's not even breaking a sweat.

I'm not the type to recommend buying ram unless it's warranted.

Yep, I've seen that exact video. The problem is in atleast a few of those games, such as AC Odyssey, I DO get max GPU usage.
That may be so, but don't you want more performance than you currently have for as least an amount/cost as possible? Getting a 2 x 8, or 2 x 16 matched kit, is gonna give you some room for menouver and plenty of bandwitdh for your CPU to feed off of.

Regarding the HDD. Whether it's an SSD or HDD, most of the time, they feed data to the CPU/Ram and their job is done. It just takes longer with a HDD to load levels. A HDD won't limit or affect your in game FPS as all the assests are first loaded to CPU/Ram and stay there.
 
Last edited:

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
"That may be so, but don't you want more performance than you currently have for as least an amount/cost as possible? "

If I knew it would give a boost, I would. The problem is half the internet things one way, half thinks the other. GamersNexus says Single Channel can actually be better in some games or marginally worse in others, never much of a difference either way. Random YouTube videos can show 50% boosts. I don't know what to go by.

"Note, I've not mentioned GPU usage in my post once. This is almost a misleading metric. "

Yes, but it's the symptom I am going by, and therefore I mentioned it, since without low GPU usage, I have no issues. Even in games that claim to need dual channel (AC Odyssey).
 
If I knew it would give a boost, I would. The problem is half the internet things one way, half thinks the other. GamersNexus says Single Channel can actually be better in some games or marginally worse in others, never much of a difference either way. Random YouTube videos can show 50% boosts. I don't know what to go by.
Well, there's not much we can offer to assuage your concerns. On the otherhand, there are many experience hardware enuthisast on this forum, and I'm practically sure 95% of them would advise the same as we are here. What you do with that is up to you. Only you can make that call. With that said, and my own experience and various hardware sites that cover this type of stuff, endless vid on this exact subject, more often than not there's a general consensus that dual channel gives more to a system than single channel can.


Yes, but it's the symptom I am going by, and therefore I mentioned it, since without low GPU usage, I have no issues. Even in games that claim to need dual channel (AC Odyssey).
Well, if that's so, just be happy that it plays as you like it to, and don't get hung up too much on 'bottlenecking' terms or such.

But to finalise, yes, I'm still of the same opinion, get a nice fast kit of ram, and be pleasantly surprised with the beneifts.
 

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
Alright, I've decided to do it. So I have the choice of 16GB DDR4 2666 for $56 or 2x16GB DDR 3200 for $119. I hear an 8700 wouldn't make much use of a 3200. The motherboard is a Z370 so it suppots up to 3200 but, again, with 8700 probably not worth replacing the stick I already have for ~$60. Opinions?
 
Last edited:

geofelt

Titan
Ram is sold in kits for a reason.
A motherboard must manage all the ram using the same specs of voltage, cas and speed.
The internal workings are designed for the capacity of the kit.
Ram from the same vendor and part number can be made up of differing manufacturing components over time.
Some motherboards, can be very sensitive to this.
This is more difficult when more sticks are involved.

If you do buy more disparate sticks, they should be the same speed, voltage and cas numbers.
Even then your chances of working are less than 100%
I might guess 90% success for intel.
What is your plan "B" if the new stick does not work?

I think I would pick the 32gb option.
I doubt you would notice any big difference in performance because of the increase to 32gb. There will be a benefit to dual channel.
But it can't hurt.
Ram performance is not only dependent on the ram speed, but also the cas timings. Lower cas is better at a given speed.
What is the exact make/model of your current ram(cpu-Z can tell you) and what of the proposed ram?
 
Reactions: Roland Of Gilead
Hello everyone!

I just upgraded to an RTX 3070 and performance is great. However, there are several games where I do not get 99% GPU usage. I'm not sure whether it's a CPU bottleneck or a RAM bottleneck, but it seems to be worse in older games where single core performance is more important, but also, my 8700K should be doing better with them. So I am considering whether or not my single channel DDR4 2666 would benefit from another stick so that I can be dual channel. The internet is incredibly bipolar about this. YouTube videos will show 20% - 40% where there's a RAM bottleneck while some articles will have charts and graphs proving that dual channel has no benefit?

Can someone help me out? I know my setup is old, 2666 is slow by today's standards, and single channel may be more of an issue now.

Also, while I'm got you guys here, any chance it's a hard drive bottleneck? That mostly shows up as stuttering right? Because I ran a lot of the games from 2015-2019 from an external HDD with 150 - 170 MBps because 10TB externals are so cheap.
Yes dual channel will give you a bump.

How much of a bump might depend on the apps you run.

It certainly won't be slower.
 
Reactions: CountMike
Alright, I've decided to do it. So I have the choice of 16GB DDR4 2666 for $56 or 2x16GB DDR 3200 for $119. I hear an 8700 wouldn't make much use of a 3200. The motherboard is a Z370 so it suppots up to 3200 but, again, with 8700 probably not worth replacing the stick I already have for ~$60. Opinions?
The IMC on the 8700k should be fine for 3200mhz. I'd go for the faster ram, with a 32gb setup ( 2 x16gb). $119 for 32gb is really good. CL16 or CL14 if you can get them. Tighter timings, maximised performance.

Damn, I iwsh we get those deals in Ireland!!! 😢
 

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
"What is your plan "B" if the new stick does not work? "

I am getting them from Crucial which is guaranteed to work with my prebuilt. If it doesn't, I return.
 

Zerk2012

Titan
Ambassador

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
Looks just like it and that's my prebuilt model number, so I assume so.

I think I can get memory cheaper from NewEgg with the labor day sales so I'm looking for that. Any comments on compatible? I assume any major brand DDR4 2666 / 3200 should work, right?
 

Zerk2012

Titan
Ambassador
Looks just like it and that's my prebuilt model number, so I assume so.

I think I can get memory cheaper from NewEgg with the labor day sales so I'm looking for that. Any comments on compatible? I assume any major brand DDR4 2666 / 3200 should work, right?
The motherboard does not support anything higher than 2666 so anything you buy will be limited to that.
 

Zerk2012

Titan
Ambassador
For your problem I'm not convinced you have a problem.

A processor can put out X amount of FPS in a given game.
Memory speed and amount can very that to a extent.
Temperatures can affect that from the motherboard or processor.

If your processor can give 100FPS in a given game and your video card can give 200FPS in that game using your settings and monitor resolution then you get close to 50% card use because it's all the rest of your system can do.

I would never use a external drive to play games.

Run this and post a link to the results.



EDIT Because 2 games were made by the same company doesn't mean they used the same game engine, version of the game engine, or the same optimizations were made.

For the results of using dual channel memory if it's available I would always use it. All the internet results say different things because of different situations can change the results. You can skew the results by using different examples to make the results go anyway you wish.
 
Last edited:

Karadjgne

Titan
Ambassador
However, there are several games where I do not get 99% GPU usage.
You don't want 99% usage, ever.
Usage is Not how much of the gpu is Used, but how much it Uses. There's a difference.

Think of it like using a hammer to put a nail in the wall. You'll use 100% of the muscles in your hand to hold the hammer, use 100% of the muscles in your arm to swing the hammer, but that's amount of muscles used, not how much they are used. You don't need 100% of the strength in your hand to hold the hammer, don't need to swing it with all the strength in your arm either.

That's usage. How much strength is brought to bear, not how many muscles are used. Better to have closer to 50-60% usage, so when you hit a wood stud, you can swing harder. Better to have that lower % so when there's explosions or smoke or other intense graphics, it doesn't drop the fps by huge amounts because there's extra strength in reserve.

Your card won't run any slower with lower usage, but it will run slower if usage is so high, it's got nothing left to give.
 
Reactions: CountMike

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
https://www.userbenchmark.com/UserRun/55081222

Here are my benchmark results. Apparently my RAM is crap, but my externals aren't much slower than my internal, and my SSD isn't as fast as I had assumed. Also... Not sure why my 3070 is in the 10th percentile ....

"Your card won't run any slower with lower usage, but it will run slower if usage is so high, it's got nothing left to give. "

Okay, so maybe now we're getting somewhere. But, I am assuming it is a sign of an issue since GPU usage is so low compared to benchmark videos and videos which demonstrate the difference between dual and single channel memory tend to have the GPU usage go up. For me, GPU usage is lower in older games that are probably more single core threaded so I am conflicted, but higher in games that are more multi-threaded but also more graphically intense so I'm going to be GPU limited anyways instead of CPU/RAM so that doesn't mean much. Anyone think they can make sense of the UserBenchmark for this question?

"Because 2 games were made by the same company doesn't mean they used the same game engine, version of the game engine, or the same optimizations were made. "

Of course, but I'm giving examples as I played them. I used these since in the last few days I have played, Control, Quantum Break, Alan Wake Remastered in that order.

Assuming I do need an upgrade,I'm currently thinking of gettingeither :

CORSAIR Vengeance LPX 32GB (2 x 16GB) 288-Pin PC RAM DDR4 2666 (PC4 21300) Intel XMP 2.0 CL16
HyperX FURY 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 2400 (PC4 19200) Desktop Memory CL15

I'm leaning towards the HyperX Fury since that is the same brand as what came with my prebuilt, so maybe it has a higher chance of going smoothly...

Anything else I need to check as far as compatibility? Probably don't need 32GB but seems weird to buy RAM in 2022 and not spend the additional $35 to get 32GB.
 
Last edited:
https://www.userbenchmark.com/UserRun/55081222

Here are my benchmark results. Apparently my RAM is crap, but my externals aren't much slower than my internal, and my SSD isn't as fast as I had assumed. Also... Not sure why my 3070 is in the 10th percentile ....

"Your card won't run any slower with lower usage, but it will run slower if usage is so high, it's got nothing left to give. "

Okay, so maybe now we're getting somewhere. But, I am assuming it is a sign of an issue since GPU usage is so low compared to benchmark videos and videos which demonstrate the difference between dual and single channel memory tend to have the GPU usage go up. For me, GPU usage is lower in older games that are probably more single core threaded so I am conflicted, but higher in games that are more multi-threaded but also more graphically intense so I'm going to be GPU limited anyways instead of CPU/RAM so that doesn't mean much. Anyone think they can make sense of the UserBenchmark for this question?

"Because 2 games were made by the same company doesn't mean they used the same game engine, version of the game engine, or the same optimizations were made. "

Of course, but I'm giving examples as I played them. I used these since in the last few days I have played, Control, Quantum Break, Alan Wake Remastered in that order.

Assuming I do need an upgrade,I'm currently thinking of gettingeither :

CORSAIR Vengeance LPX 32GB (2 x 16GB) 288-Pin PC RAM DDR4 2666 (PC4 21300) Intel XMP 2.0 CL16
HyperX FURY 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 2400 (PC4 19200) Desktop Memory CL15

I'm leaning towards the HyperX Fury since that is the same brand as what came with my prebuilt, so maybe it has a higher chance of going smoothly...

Anything else I need to check as far as compatibility? Probably don't need 32GB but seems weird to buy RAM in 2022 and not spend the additional $35 to get 32GB.
I think you will find that the fastest ram your mobo will support is 2666.
 

Zerk2012

Titan
Ambassador
https://www.userbenchmark.com/UserRun/55081222

Here are my benchmark results. Apparently my RAM is crap, but my externals aren't much slower than my internal, and my SSD isn't as fast as I had assumed. Also... Not sure why my 3070 is in the 10th percentile ....

"Your card won't run any slower with lower usage, but it will run slower if usage is so high, it's got nothing left to give. "

Okay, so maybe now we're getting somewhere. But, I am assuming it is a sign of an issue since GPU usage is so low compared to benchmark videos and videos which demonstrate the difference between dual and single channel memory tend to have the GPU usage go up. For me, GPU usage is lower in older games that are probably more single core threaded so I am conflicted, but higher in games that are more multi-threaded but also more graphically intense so I'm going to be GPU limited anyways instead of CPU/RAM so that doesn't mean much. Anyone think they can make sense of the UserBenchmark for this question?

"Because 2 games were made by the same company doesn't mean they used the same game engine, version of the game engine, or the same optimizations were made. "

Of course, but I'm giving examples as I played them. I used these since in the last few days I have played, Control, Quantum Break, Alan Wake Remastered in that order.

Assuming I do need an upgrade,I'm currently thinking of gettingeither :

CORSAIR Vengeance LPX 32GB (2 x 16GB) 288-Pin PC RAM DDR4 2666 (PC4 21300) Intel XMP 2.0 CL16
HyperX FURY 32GB (2 x 16GB) 288-Pin DDR4 SDRAM DDR4 2400 (PC4 19200) Desktop Memory CL15

Anything else I need to check as far as compatibility? Probably don't need 32GB but seems weird to buy RAM in 2022 and not spend the additional $35 to get 32GB.
The dual channel will make the most difference but a few things.

Your SSD is on the edge of to full about 90% of any drive is all you really need to use. Any of them that are 2666 are good options.

For the Crucial 100% guaranteed to work glad you got away from that.
I checked everything with a fine tooth comb a couple years ago.
Nowhere on the site did they say they guarantee the memory you buy to work with your excising memory only that it is compatible with your PC.

For the video card download and look at GPUZ and make sure it's running @ the right speed,16 x 3.0 I think with your board.
 

Chris8282

Distinguished
Mar 1, 2013
41
1
18,535
0
Alright, then, so is CL15 worth $5 for dollars than CL16?

"For the video card download and look at GPUZ and make sure it's running @ the right speed,16 x 3.0 I think with your board. "

I've got GPU-Z where do I find this information. i.e. what is it called? Bus interface? It says 'PCIe x 16 4.0 @ 16 1.1
 
Last edited:

Karadjgne

Titan
Ambassador
Your benchmark isn't very accurate. You have too much going on in the background. Close out all that stuff and then run it.

Clean out the ssd. You've got way too much in it for any use case and all kinds of storage space elsewhere. Windows needs space to work and you are crimpjng it tight, which Will slow down the ssd. It's a 256Gb Toshiba, you'll want to not be much
more than 50-60% capacity at any given time.

You aren't going to see any accurate results until you clean up the pc and actually get it running right. If there's any issues they'll be self evident.

Can't complain about a car that's running and driving badly if you won't put air in the tires, fix the bad shocks and at least throw in some fuel injector cleaner.
 
Reactions: CountMike

ASK THE COMMUNITY