[SOLVED] RTX 3060 not performing well

[Huema838]

Hey, so I've recently upgraded from a GTX 1070 to a RTX 3060. I thought the upgrade would improve performance greatly, but it hasn't. Games constantly fluctuate in fps and stutter now, and even in CS:GO, an easy game to run, I see only 200 fps even while on a lower resolution.

My specs
GPU: RTX 3060
CPU: i7-7700k
RAM: 24gb DDR4 (16g and 8gb Crucial)
Motherboard: GA-H270-HD3

I've tried DDU and using Malaware Bytes, but nothing seems to be working and I'm unsure of what the next steps are.
Any help would be well appreciated.

 

[FoxInFlames]

Not sure, might be bottleneck, do you get 200 fps even at higher resolution and settings?

 

[Huema838]

I've run a UserBenchmark to add:

Gigabyte GA-H270-HD3-CF Performance Results - UserBenchmark
UserBenchmarks: Game 82%, Desk 85%, Work 72%
CPU: Intel Core i7-7700K - 83.6%
GPU: Nvidia GeForce RTX 3060 - 99.7%
SSD: Crucial M550 128GB - 62.8%
HDD: Seagate Barracuda 1TB (2016) - 76.9%
RAM: Unknown CT16G4DFD8266.C16FE 859B CT8G4DFD8213.C16FBD2 24GB - 66.9%
MBD: Gigabyte GA-H270-HD3-CF

 

[FoxInFlames]

Well, benchmarks just tell how good it is, not if it's as good as it's supposed to be.

Just tell if you get 200 fps even after improving the graphics.

 

[Huema838]

Not sure, might be bottleneck, do you get 200 fps even at higher resolution and settings?

The fps improves only decreases by 10 - 15 FPS no matter what graphical settings I use

 

[FoxInFlames]

Seems bottlenecking indeed, might want to consider upgrading your cpu, atleast from my POV

 

[dotas1]

I've run a UserBenchmark to add:

Gigabyte GA-H270-HD3-CF Performance Results - UserBenchmark
UserBenchmarks: Game 82%, Desk 85%, Work 72%
CPU: Intel Core i7-7700K - 83.6%
GPU: Nvidia GeForce RTX 3060 - 99.7%
SSD: Crucial M550 128GB - 62.8%
HDD: Seagate Barracuda 1TB (2016) - 76.9%
RAM: Unknown CT16G4DFD8266.C16FE 859B CT8G4DFD8213.C16FBD2 24GB - 66.9%
MBD: Gigabyte GA-H270-HD3-CF

Multiple issues detected here.
OS drive is almost full which even at 40% free space you get gradually major performance drop in those SSDs. Imagine with 10% free.

The mix or ram is very much likely to contribute to further performance drop. Usually you need a kit of ram being sold together because they are tested for compatibility and you are certain they play along perfectly which is not what generally happens when you mix and match. The speed of ram is another issue since you only have the base speed that makes the system slower.

The CPU does not boost to it's potential so you may have a thermal issue as well or it can be from the rest of the issues mentioned above.

What is the make and model of your PSU?

 

[FoxInFlames]

OS drive is almost full which even at 40% free space you get major performance drop in those SSDs. Imagine with 10% free.

Really? I have 55 gigs free out of 240 (around 23%) yet my M.2 did get to the 62nd percentile with a 110% (it's a Kingston)

I use two different hynix sticks, one of 2666 and other of 2400MHz yet i reached the 100th percentile with a 79%

My i5 9300H came through with 84th (81.9%) and my 1650 mobile reached 95th (45.5%)

Couldn't check my hdd though, ram cached (optane memory, 16 gigs)

I stand by what i said, the weakest link here is the cpu, the 3060 demands atleast an i5 12600k for minimal bottleneck, which is 25% faster than the 7700k

My benchmark:- https://www.userbenchmark.com/UserRun/49308034

 

[dotas1]

Really? I have 55 gigs free out of 240 (around 23%) yet my M.2 did get to the 62nd percentile with a 110% (it's a Kingston)

I use two different hynix sticks, one of 2666 and other of 2400MHz yet i reached the 100th percentile with a 79%

My i5 9300H came through with 84th (81.9%) and my 1650 mobile reached 95th (45.5%)

Couldn't check my hdd though, ram cached (optane memory, 16 gigs)

I stand by what i said, the weakest link here is the cpu, the 3060 demands atleast an i5 12600k for minimal bottleneck, which is 25% faster than the 7700k

My benchmark:- https://www.userbenchmark.com/UserRun/49308034

Depending on the SSD's technology, if it's DRAMless and it's capacity, you get ofcourse different results. Small capacities tend to slow down much more than bigger capacities for example.

Your RAM will both run at 2400MHz and mixing does not mean that they will not work, it means that you might bump into problems instead.

Userbenchmark is NOT to be taken seriously as it has many flaws and takes account even the highly and heavy OCd systems and put them in the same bag with the non OCd. Also people with major OC tend to run the benchmark again and again which makes the mean rise. It really helps identifying issues with hardware if you know where to look.

In higher resolutions the CPU is becoming less important and since he uses a 1440p monitor it's more than enough for now.

 

[John Chesterfield]

Depending on the SSD's technology, if it's DRAMless and it's capacity, you get ofcourse different results. Small capacities tend to slow down much more than bigger capacities for example.

Your RAM will both run at 2400MHz and mixing does not mean that they will not work, it means that you might bump into problems instead.

Userbenchmark is NOT to be taken seriously as it has many flaws and takes account even the highly and heavy OCd systems and put them in the same bag with the non OCd. Also people with major OC tend to run the benchmark again and again which makes the mean rise. It really helps identifying issues with hardware if you know where to look.

In higher resolutions the CPU is becoming less important and since he uses a 1440p monitor it's more than enough for now.

Right on the money.

 

[Huema838]

Multiple issues detected here.
OS drive is almost full which even at 40% free space you get gradually major performance drop in those SSDs. Imagine with 10% free.

The mix or ram is very much likely to contribute to further performance drop. Usually you need a kit of ram being sold together because they are tested for compatibility and you are certain they play along perfectly which is not what generally happens when you mix and match. The speed of ram is another issue since you only have the base speed that makes the system slower.

The CPU does not boost to it's potential so you may have a thermal issue as well or it can be from the rest of the issues mentioned above.

What is the make and model of your PSU?

Segotep GP-700G 600Whttps://pricespy.co.nz/product.php?e=3599440

 

[dotas1]

Segotep GP-700G 600Whttps://pricespy.co.nz/product.php?e=3599440

That's a bit on the low side of mediocre PSUs. I wouldn't keep using it for long if I were you.

First things first.
How are your temps of CPU and GPU idle and while gaming? Use hwinfo sensors mode or MSI afterburner to find out.

Free some SSD space if possible by moving some things to the HDD, or even better buy another bigger capacity SSD to install OS there instead (if you can afford of course).

Scan system for malware with Malwarebytes.

Next steps would be selling the ram you own and buying a dual kit of RAM with higher speed, something like 3000- 3200MHz which are currently rather cheap.

The last thing I will mention is the first I said. It's better to buy a quality PSU. It is what powers everything and if the power delivered to your hardware is not clean or adequate, problems occur. It affects the longetivity and performance of your system. I would suggest tier B or above from this tier list:

Article - [psucultists] PSU tier list rev. 14.8 (Final Update Jul '21)

PSU Tier List 4.0 rev. 14.8 (outdated) Last Update: 12-07-2021 Legend : Gray - EoL/obsolete and/or otherwise not recommended for purchase. Green - small form-factor (gold and blue colors are disregarded due to scarcity of SFX PSUs) Gold - best units in the tier (includes requirements for blue...

forums.tomshardware.com

 

[Huema838]

That's a bit on the low side of mediocre PSUs. I wouldn't keep using it for long if I were you.

First things first.
How are your temps of CPU and GPU idle and while gaming? Use hwinfo sensors mode or MSI afterburner to find out.

Free some SSD space if possible by moving some things to the HDD, or even better buy another bigger capacity SSD to install OS there instead (if you can afford of course).

Scan system for malware with Malwarebytes.

Next steps would be selling the ram you own and buying a dual kit of RAM with higher speed, something like 3000- 3200MHz which are currently rather cheap.

The last thing I will mention is the first I said. It's better to buy a quality PSU. It is what powers everything and if the power delivered to your hardware is not clean or adequate, problems occur. It affects the longetivity and performance of your system. I would suggest tier B or above from this tier list:

Article - [psucultists] PSU tier list rev. 14.8 (Final Update Jul '21)

PSU Tier List 4.0 rev. 14.8 (outdated) Last Update: 12-07-2021 Legend : Gray - EoL/obsolete and/or otherwise not recommended for purchase. Green - small form-factor (gold and blue colors are disregarded due to scarcity of SFX PSUs) Gold - best units in the tier (includes requirements for blue...

forums.tomshardware.com

Hey, I just got 16gbs of 3200mhz ram, but when installing it I realised that my motherboard only supports 2400 mhz and I don't know If i can overclock it past that. FPS spikes and drops are still happening along with the unplayable stuttering.

Here are the temps anyway -
GPU Idle: 53° C
CPU Idle: 46 - 50° C

Gaming Temps -
GPU: 65° C
CPU:68 - 72° C

I've done the malwarebytes thing you recommended but that didn't help. I now have 40 gbs free on my ssd, but nothing is seeming to help. Im so lost on what to do, I thought the ram would fix it but It hasn't helped.

 

[dotas1]

Hey, I just got 16gbs of 3200mhz ram, but when installing it I realised that my motherboard only supports 2400 mhz and I don't know If i can overclock it past that. FPS spikes and drops are still happening along with the unplayable stuttering.

Here are the temps anyway -
GPU Idle: 53° C
CPU Idle: 46 - 50° C

Gaming Temps -
GPU: 65° C
CPU:68 - 72° C

I've done the malwarebytes thing you recommended but that didn't help. I now have 40 gbs free on my ssd, but nothing is seeming to help. Im so lost on what to do, I thought the ram would fix it but It hasn't helped.

Yes, I just saw that you have a h270 motherboard. It has limits to RAM clocks and ofcourse no overclocking of the 7700K. I am not very sure but I think you can't OC beyond 2400. That's the issue with H series motherboards, no OC is allowed. The RAM you bought should help with all this but probably a bigger issue does not let you see the benefits yet.

Are you sure you are reporting correct temps? Something is keeping your CPU from boosting to it's advertised 4,5GHz and usually that's thermal issues. Do you have the latest BIOS, chipset drivers, windows updates?

I guess you are gaming through HDD right? That does slow things down and might be responsible for SOME of the drops and stuttering but I don't think it's all that, unless you have a failing HDD (or even SSD). Run crystalmark and check drives health status.

 

[jasonf2]

You are running a 4 core Kaby lake processor circa 2017. Game pre-render is ran on CPU and is usually the real determining factor of Frame rate. GPU is the determinate for fidelity settings. I don't like to really use the term bottleneck, but anything with less than 6 cores and near current generation IPC is underpowered to run a 3060 to full potential. That is why changing settings makes almost no difference in end frame rate. If you put a 3090 on that chip I would expect similar results.

 

[WarWolverineWarrior]

What Jasonf2 said, you're not going to see real change in 1080p fps with 1070 to 3060, especially in CS:GO. The CPU is the limiting factor in 1080p. With 3060, you're going to see very little FPS difference in Low or Ultra settings at 1080p. 1080p is more CPU bound while 4k is more GPU bound. You're better off buying a 1440p monitor to see the improvement of getting this GPU.
You should also only run 1,2 or 4 sticks of same RAM.

 

[helper800]

You are running a 4 core Kaby lake processor circa 2017. Game pre-render is ran on CPU and is usually the real determining factor of Frame rate. GPU is the determinate for fidelity settings. I don't like to really use the term bottleneck, but anything with less than 6 cores and near current generation IPC is underpowered to run a 3060 to full potential. That is why changing settings makes almost no difference in end frame rate. If you put a 3090 on that chip I would expect similar results.

That CPU is fine for 1440p. My friend has a 4670k and a 3060 ti and gets 310 fps in CS:GO @ 1080p. Clearly something is going on with his CPU other than its "age."

 

[John Chesterfield]

That CPU is fine for 1440p. My friend has a 4670k and a 3060 ti and gets 310 fps in CS:GO @ 1080p. Clearly something is going on with his CPU other than its "age."

Yeah, something not right with software somewhere.

 

[helper800]

What Jasonf2 said, you're not going to see real change in 1080p fps with 1070 to 3060, especially in CS:GO. The CPU is the limiting factor in 1080p. With 3060, you're going to see very little FPS difference in Low or Ultra settings at 1080p. 1080p is more CPU bound while 4k is more GPU bound. You're better off buying a 1440p monitor to see the improvement of getting this GPU.
You should also only run 1,2 or 4 sticks of same RAM.

He is on a 1440p monitor.

 

[jasonf2]

That CPU is fine for 1440p. My friend has a 4670k and a 3060 ti and gets 310 fps in CS:GO @ 1080p. Clearly something is going on with his CPU other than its "age."

There is a 77% increase in pixel count between 1440p and 1080p. Also saying that I have a friend who hit 300 fps at 1080 on an older CPU, not the CPU in question, has absolutely no relevance to this persons machine. It does perhaps establish a baseline that you can hit 300 fps at 1080 on a 4 core intel chip but not much more. The end user is still getting 200+ fps @ 1440. Nothing described here is unusual when older machines with decent GPUs (for their time) throw a current gen GPU in and expect massive frame rate increase.

In the point of good argument most machines can be tweaked and upgraded to get better performance. How much here I cannot say. My best guess is if the rust drive is being used some load time could be reduced by moving over to a complete SSD rig. The end user has mix and match RAM on uneven stick configs. There might be a little to get back there with a solid memory kit at XMP and someone that knows what they are doing might be able to eak a few extra mhz out of the 7700k. Is it going to hit 300fps though, probably not.

 

[helper800]

There is a 77% increase in pixel count between 1440p and 1080p. Also saying that I have a friend who hit 300 fps at 1080 on an older CPU, not the CPU in question, has absolutely no relevance to this persons machine. It does perhaps establish a baseline that you can hit 300 fps at 1080 on a 4 core intel chip but not much more. The end user is still getting 200+ fps @ 1440. Nothing described here is unusual when older machines with decent GPUs (for their time) throw a current gen GPU in and expect massive frame rate increase.

In the point of good argument most machines can be tweaked and upgraded to get better performance. How much here I cannot say. My best guess is if the rust drive is being used some load time could be reduced by moving over to a complete SSD rig. The end user has mix and match RAM on uneven stick configs. There might be a little to get back there with a solid memory kit at XMP and someone that knows what they are doing might be able to eak a few extra mhz out of the 7700k. Is it going to hit 300fps though, probably not.

The higher the resolution the less CPU dependent a game becomes, however, CPUs still determine maximum FPS in cases where the GPU is much more powerful than the CPU. The pixel count does nothing but benefit older CPUs because it, again, shifts the workload onus to the GPU. I have confirmed through benchmarking of other publications that with his system he should be getting much more FPS with a 7700k and 3060 pairing in CS:GO (360 fps ultra at 1080p, and 295fps ultra 1440p). If you looked at the userbenchmark result he posted clearly there is something going on with the performance of his CPU (the maximum FPS limiter)

 

[jasonf2]

The higher the resolution the less CPU dependent a game becomes, however, CPUs still determine maximum FPS in cases where the GPU is much more powerful than the CPU. The pixel count does nothing but benefit older CPUs because it, again, shifts the workload onus to the GPU. I have confirmed through benchmarking of other publications that with his system he should be getting much more FPS with a 7700k and 3060 pairing in CS:GO (360 fps ultra at 1080p, and 295fps ultra 1440p). If you looked at the userbenchmark result he posted clearly there is something going on with the performance of his CPU (the maximum FPS limiter)
[/QUOTE}
My personal experience with userbenchmark is mixed. It doesn't take everything into account and while it maybe represents a best case scenario there is a lot of other stuff going on in homebuilt machines or machines with aftermarket parts . Like in this case the mixed ram on uneven stick count (There is a reason it is called DDR) and a rust drive in the system. Neither are a real problem per se (as long as the machine is stable), but both will contribute to the CPU being fed just a little less than they could be. That isn't even starting with overclocks and cooling solutions. At the end of the day maximum framerate is dependent on CPU throughput and then reduced by the ability of the GPU to tack on fidelity. If the end user hasn't done a full windows reinstall in a while will they see a performance increase as well? Probably. Would I suggest doing it on a stable build? Not a chance unless the end user is fully aware of the consequences, and certainly not with an expectation that this machine will start cranking out 290fps.