[SOLVED] RX 580 and i7-3770 Help.

[gagecarmongaming]

Alright so an issue I'm having is that I am getting seemingly BIG stutters in games and performance seems to be below par of videos I've found online of similar builds. I am not the best PC builder but I've learned a few things over the years, though not enough to figure out what's going on with this PC! Here are my specs for starters.

GPU - MSI RX 580 8GB OC Model.
CPU - i7-3770 Intel Processor
RAM - DDR3 Patriot Viper 2 x 8gb 1600mhz in Dual Channel Mode
Storage - 250GB Samsung 860 Evo SSD (OS installed on here. Windows 10 Home) and a 1TB Seagate BarraCuda 6gb/s 64mb Cache
Motherboard - MSI H61MU-E35(B3)(MS-7680)
PSU - EVGA 750watt Power Supply

AMD states that I'm getting on average around 105 fps in Fortnite and 196.3 in TF2 for example, but the frames dip REALLY hard. I thought that this may be a frametime issue with the spikes I see on the graph when using MSI Afterburner's OSD. My current monitor is a little 60hz Proscan TV which is getting replaced by a 144hz Asus monitor, not sure if a monitor can cause these issues but stating what I have is better than nothing!

(I'm planning on trying to move onto a AMD CPU build whenever I can get the cash collected for the parts, but I really want to get this fixed forcurrent use since I may be giving this one to my younger brother later down the road.)

 

[4745454b]

3770 is old. How is the CPU and board? Temps? Are you sure it's a 3770 and not 3750? And it's not a K? I'd check the temps and clock speeds of the CPU and GPU.

 

[gagecarmongaming]

It's a 3770 for sure, not a 3770k. The temps for the GPU and CPU usually stay below 70 but the temp of the GPU can rise pretty fast on certain games like Metro Last Light and Generation Zero. I'm surprised my 3770 has lasted this long for how old it is, I've been using it in VR and expected much less from it. It's certainly a fighter.

Playing Fortnite at high settings has my GPU at 64% and bounces around, while also running 1146mhz memory usage while also bouncing. CPU usage is at 90% and bouncing while at 3392mhz for the clock on it.

The CPU when looking at it physically seems really great, same for the motherboard. Though I ordered these a few years ago used. The only new components are the CPU cooler, SSD and HDD, then the RAM.

It seems to sometimes keep steady but then spikes over and over. One thing I've noticed is it sometimes does it while there's a lot of the world in view.

 

[4745454b]

Does the CPU stay at 3.3GHz? 3.4GHz? Any temp sensors way off? Might be an issue with the board.

 

[gagecarmongaming]

The i7 stays at 3.392ghz inside and outside of games from what MSI is showing. I haven't noticed any temperatures going insane on any of the threads.

Edit: Just went into a game to test it. I had large lag spikes as the CPU was reaching 90% usage. I then lowered all graphics settings to Low and had the same issue. So I assume it's a CPU issue or motherboard issue?

Another edit!: I have noticed while trying to stress my PC out on Halo Infinite that my GPU is only drawing 80 watts of power while the CPU bounces from 60% used to 90% and all over. I'm now wondering if it's a issue of the motherboard not distributing power correctly.

 

[gagecarmongaming]

Okay so I've been reading up on my specific model of GPU and on the official MSI website it states that the GPU core clock can get up to 1366mhz. While under a stress test, mine only reaches 1200mhz and uses 120watts of power. Interesting enough, both the wattage and core clock drop after a mere few seconds of being used. It then does it's usually rising and dropping while leaving my PC slow a few seconds later. Video memory seems to show fine. It reads 2180mhz on Afterburner, but I read that you have to multiply that by 4 to find the real memory clock readings which is around 8720. Which from MSI's spec page is 720mhz too high.

Edit: I found one of the issues! I checked the VBIOS of my GPU and it lead back to a different version of the RX580. I then flashed the correct BIOS onto my GPU and now it's reaching the 1366 core clock and 8000mhz memory clock! Only thing I can assume is that maybe the previous owner was using it for mining.

 

[4745454b]

That's possible. Miners can flash or change the bios so it hits different speeds. Now that you are hitting 1366MHz are you getting better frame rates?

 

[gagecarmongaming]

That's possible. Miners can flash or change the bios so it hits different speeds. Now that you are hitting 1366MHz are you getting better frame rates?

I am though the spikes are still relevant. I also have noticed that the GPU is getting hot really fast. I'm reaching 120fps in main menu but temps are reaching past 85 degrees celsius. I just replaced the thermal paste today to see if that would help but no luck. I'm going to try leaving the side panel off for a bit to see if the case needs more cool air. Will edit this with results.

Edit: With the panel off, it stays around 80 to 83 degrees. GPU usage seems normal but then goes to 20 or 30 as the lag spikes kick in. CPU usage seems eratic as well. I have no idea what to do.

 

[4745454b]

While I don't like or let my CPU get very hot, GPUs are another matter. 80 or 85C isn't out of line for a GPU. Some are known to hit 105C with heavy use. 80C on a CPU I'd say you have an issue. 80C on the GPU is normal. So if your frame rates are good and everything is working fine and you are hitting 85C on your GPU I'd argue to leave things alone.

Spikes could be anything. Could be something throttling down because of heat, or something loading from a slow spinning drive, or things needing to be cleared from RAM so new can be loaded, etc. If the GPU usage seems normal but goes to 30% load when the lag spike hits, you need to figure out what's happening when that occurs. Is it reading from the drive? Is the CPU throttling down? Is the motherboard getting to hot and dropping the CPU speed? Is the ram flushing? Your doing a great job with looking.

Something to consider, the 580 is an older card and if it was being mined on than who knows what kind of life it still has. While you finally got your 1366MHz core speed, you might want to try lowering it to 1300, or 1275? You can probably drop the GPU Vcore a bit as well. Should help with temps. IF its a GPU issue this might help.

 

[gagecarmongaming]

While I don't like or let my CPU get very hot, GPUs are another matter. 80 or 85C isn't out of line for a GPU. Some are known to hit 105C with heavy use. 80C on a CPU I'd say you have an issue. 80C on the GPU is normal. So if your frame rates are good and everything is working fine and you are hitting 85C on your GPU I'd argue to leave things alone.

Spikes could be anything. Could be something throttling down because of heat, or something loading from a slow spinning drive, or things needing to be cleared from RAM so new can be loaded, etc. If the GPU usage seems normal but goes to 30% load when the lag spike hits, you need to figure out what's happening when that occurs. Is it reading from the drive? Is the CPU throttling down? Is the motherboard getting to hot and dropping the CPU speed? Is the ram flushing? Your doing a great job with looking.

Something to consider, the 580 is an older card and if it was being mined on than who knows what kind of life it still has. While you finally got your 1366MHz core speed, you might want to try lowering it to 1300, or 1275? You can probably drop the GPU Vcore a bit as well. Should help with temps. IF its a GPU issue this might help.

The CPU temps seem to be perfectly fine, even when it reaches 100% usage while gaming. It only ever reaches around 70 celsius. With the GPU however, the fans get insanely loud so I've been trying to keep them under 1900 RPM which seems to be barely enough to keep the GPU just at 85 celsius. Well seeing as how it's large spikes in a few titles and not others, I'm assuming it could potentially just be the game being unoptimized for my PC.

When the large spikes happen, both the CPU and GPU drop their usage at the same time if that helps with narrowing down what it could be. I've actually been getting notifications before that my GPU is getting too hot at 75 celsius, haven't even seen if my GPU can hit the 90 celsius advisor point.

Thank you for complimenting me on trying to diagnose the problem, I've never tried this before so never really knew what to look for. Keeps my confidence up in finding the solution to my PC's problems!


I'll try seeing if the RAM is flushing out and seeing if the motherboard is getting too hot as you advised.

 

[lordmogul]

With the CPU and GPU loads you are reporting, even with the lower GPU clock, it sounds like a CPU limit.

Temps on the CPU aren't too surprising, considering there is just thermal paste under the heatspreader, and not a really great one. On my old 3570K I could drop temps by around 10-15 °C by delidding and replacing the paste with liquid metal. But any decent, fresh paste should help. Stuff can dry out after years of usage.

For the GPU, might be worth cleaning it and putting new paste on to get cooling there out of the equation as well.

Those drops in load on both, CPU and GPU, might hint at loading issues. Either with the RAM itself (might be nearly full, Windows starts paging at around 75-80% usage), or when the game tries to load stuff from disc and has to wait.

Have you tried running the usual benchmarks and compared your scores to other systems with the same parts? Should be within 5-10% depending on oc, thermals, etc.

 

[gagecarmongaming]

With the CPU and GPU loads you are reporting, even with the lower GPU clock, it sounds like a CPU limit.

Temps on the CPU aren't too surprising, considering there is just thermal paste under the heatspreader, and not a really great one. On my old 3570K I could drop temps by around 10-15 °C by delidding and replacing the paste with liquid metal. But any decent, fresh paste should help. Stuff can dry out after years of usage.

For the GPU, might be worth cleaning it and putting new paste on to get cooling there out of the equation as well.

Those drops in load on both, CPU and GPU, might hint at loading issues. Either with the RAM itself (might be nearly full, Windows starts paging at around 75-80% usage), or when the game tries to load stuff from disc and has to wait.

Have you tried running the usual benchmarks and compared your scores to other systems with the same parts? Should be within 5-10% depending on oc, thermals, etc.

The CPU cooler I'm currently using is a Coolermaster Hyper 212 Black Edition. I've changed the CPU's thermal paste a few times over the past year which seems to be fine. I just changed the thermal paste on my GPU today, still the same temps sadly. Cleaned out the fans as well.

From a screenshot I took before fixing the VBIOS of the GPU, I can see that 9gb's of my DDR3 RAM is being used while playing out of the 16gb's installed.

I have noticed that textures take a bit to load in even with the game installed on the SSD. Though the same thing happens with Halo Infinite Multiplayer installed on my HDD. When the world is loading in or if I'm traversing the world in each game, it seems to freeze for a second or so, or while turning my position in game too quickly.

Where do I benchmark my PC and compare the results? All I have is videos online to compare it to which isn't very viable or reliable.

 

[gagecarmongaming]

Here are results from Userbenchmark


UserBenchmarks: Game 42%, Desk 80%, Work 35%
CPU: Intel Core i7-3770 - 76.5%
GPU: AMD RX 580 - 51.4%
SSD: Samsung 860 Evo 250GB - 50.6%
HDD: Seagate Barracuda 1TB (2016) - 69.4%
RAM: Unknown 1600 CL9 Series 2x8GB - 55.3%
MBD: MSI H61MU-E35(B3)(MS-7680)

UserBenchmark tells me to make sure I plug my SSD into a SATA 3.0 port with a 3.0 cable and turn XMP Profile on in BIOS.

 

[4745454b]

Those drops in load on both, CPU and GPU, might hint at loading issues.

If the CPU drops, it will cause a GPU drop. So why is the CPU dropping? Heat would be the first guess. But 70C should be fine. Could be a board power issue. Or it's stalling from the RAM/drive. If the CPU has to wait for info then the effect is the same.

SSD, RAM, and GPU are all ~50%. I'm not sure how the SSD is connected or to which port. Looking online, your board lacks any SATA III ports. If its slowing because of loading there isn't much you can do. Check the ram speed. Dual channel?

 

[gagecarmongaming]

The CPU is usually around 90% usage and goes up to 100% usage. I assume maybe that's why everything is dropping? The SSD is connected to the motherboard with a SATA cable into SATA Port 1, with the HDD plugged into SATA Port 2. My RAM is in Dual channel mode and I just turned on the XMP profile in BIOS like Userbenchmark insisted. After doing another test and reinstalling Windows 10 to enable AHCI in BIOS, my SSD performance went up 10% and RAM stayed the same even with XMP enabled.


UserBenchmarks: Game 45%, Desk 80%, Work 37%
CPU: Intel Core i7-3770 - 77.6%
GPU: AMD RX 580 - 54.1%
SSD: Samsung 860 Evo 250GB - 68.4%
HDD: Seagate Barracuda 1TB (2016) - 73.4%
RAM: Unknown 1600 CL9 Series 2x8GB - 55.3%
MBD: MSI H61MU-E35(B3)(MS-7680)

 

[gagecarmongaming]

I opened up Resource Manager and saw the CPU was completely fine but noticed the Disk the game was running on froze everytime my Disk usage spiked.

Open Resource Manager screenshot

 

[4745454b]

As I posted above, your board lacks any SATA III ports. I wouldn't think that would affect performance that much, it might cause a delay in loading. More so if you have a smaller/slower SSD that's filling up.

 

[gagecarmongaming]

Wait. So is it because the motherboard doesnt have SATA 3 being the reason that it's spiking and lagging general performance, or are you meaning that it could be something else?

 

[4745454b]

https://www.msi.com/Motherboard/H61MUE35_B3/Specification

According to this you don't have any SATA III ports. So you should max out at 300MBps transfer instead of the 500-550MBps the drive is capable of. When I got my 840pro drive and did some real world testing between it and my older 470)?) drive I found nearly no difference. Some games or levels would load a single second faster, but performance was nearly the same. At least for my system at the time. Personally I think the issue is something else, but you can't plug it into a SATA III port because you don't have any.

 

[gagecarmongaming]

https://www.msi.com/Motherboard/H61MUE35_B3/Specification

According to this you don't have any SATA III ports. So you should max out at 300MBps transfer instead of the 500-550MBps the drive is capable of. When I got my 840pro drive and did some real world testing between it and my older 470)?) drive I found nearly no difference. Some games or levels would load a single second faster, but performance was nearly the same. At least for my system at the time. Personally I think the issue is something else, but you can't plug it into a SATA III port because you don't have any.

Ah alright, I understand now. I have been trying to pay attention to the RAM to make sure it isn't the issue. I have noticed that it spikes with hard faults. I have been overseeing this since I was focusing on the storage and watching in pain as the GPU usage/power consumption and CPU usage kept dropping every now and then when it lags.

 

[4745454b]

Hard faults? So the ram is bad? Have you ran memtest?

https://www.techpowerup.com/memtest64/?ref=dtf.ru

 

[gagecarmongaming]

I have not, I am doing that immediately.

 

[Bob.B]

The CPU is usually around 90% usage and goes up to 100% usage. I assume maybe that's why everything is dropping? The SSD is connected to the motherboard with a SATA cable into SATA Port 1, with the HDD plugged into SATA Port 2. My RAM is in Dual channel mode and I just turned on the XMP profile in BIOS like Userbenchmark insisted. After doing another test and reinstalling Windows 10 to enable AHCI in BIOS, my SSD performance went up 10% and RAM stayed the same even with XMP enabled.


UserBenchmarks: Game 45%, Desk 80%, Work 37%
CPU: Intel Core i7-3770 - 77.6%
GPU: AMD RX 580 - 54.1%
SSD: Samsung 860 Evo 250GB - 68.4%
HDD: Seagate Barracuda 1TB (2016) - 73.4%
RAM: Unknown 1600 CL9 Series 2x8GB - 55.3%
MBD: MSI H61MU-E35(B3)(MS-7680)

When you run UBM reboot and then wait a few mins.
Run UBM with the browser closed and post a LINK to the results page.

 

[gagecarmongaming]

Hard faults? So the ram is bad? Have you ran memtest?

https://www.techpowerup.com/memtest64/?ref=dtf.ru

I have let the application loop around 6 times over 15 minutes, no errors have been detected. I also used the Samsung Magician application earlier to test my SSD just in case, no errors were found during a diagnostic scan (even saying the SSD was working at 100% efficiency) and temps were at 40 degrees during the performance benchmark.

So with my analysis, I can assume the RAM and storage is likely not the issue?

 

[gagecarmongaming]

When you run UBM reboot and then wait a few mins.
Run UBM with the browser closed and post a LINK to the results page.

Alright, will do!