[SOLVED] Trying to diagnose bottleneck

[HeathenHeart]

PC Specs:
i7 4790k
Gigabyte RTX 2070 Windforce
2x8GB 1866MHz Ram
120GB SSD
650W PSU

So, I bought a brand new RTX 2070 Super, fired up some games and got instantly dissapointed with the performance. Playing on 1080p, I got several dips into 35fps on games that I should be getting steady 60fps according to the internet.

At first, I figured that I've got a driver issue, so I've used DDU, made a clean install, but the problem persisted.

I've benchmarked on Unigine Heaven and got a 4001 score, which seems on par with the model. Userbenchmark also gave me higher than average results.

Temperatures seem fine, both on my GPU and my CPU. Lowering graphics settings doesn't seem to help at all.

So, I'm figuring that I have a CPU bottleneck. Not surprising since my CPU is 5 years old, but i'd really like to be sure before I shell out $1000+ on a new processor and mobo.

I've logged the game's benchmark (test starts at the dash line). GPU Usage starts at about 95% at the easy part of the benchmark, and drops precipitously right when the game gets hectic. CPU Usage is consistently high on some cores throughout the whole test.

All of this screams cpu bottleneck to me, but am I missing something? Might it be something else?

Thanks!

 

[Darkbreeze]

What game?

What is the EXACT model of your 650w PSU?

DO you have the MOST recent motherboard BIOS version installed? What IS your motherboard model? Have you monitored your VRM temperatures IN ADDITION to your CPU and GPU temps?

What is your CPU cooler model, case model, number of case fans, exact locations and orientations for EACH case fan?

Do you have the latest Intel chipset driver for your motherboard chipset, (Which is?), direct from the Intel website rather than the motherboard product page, installed?

When you ran the DDU, did you run it while in Safe Mode? What was your previous card model?

Do you have the MOST recent drivers for your network adapters from your motherboard product page, directly, installed?

Do ANY other games have this problem or is it only this specific game?

 

[HeathenHeart]

The game in question is Total War: Three Kingdoms, on Ultra. But I have the same behavior on Assassin's Creed Odyssey and Total War Warhammer II, which are the other reasonably demanding games that I have installed. FPS Drops while the GPU is idle at 40-60% and the CPU is on high usage.

PSU is a XFX XT 600 (not 650, sorry).

Cooler is a Corsair H50. Case is a Corsair Spec-01, 2 fans on front, positive pressure on the case.

Mobo is ASUS Z97k. I've never updated neither the BIOS version, nor the chipset/network adapter drivers.

Previous card model was GTX 970. I ran DDU through safe mode.

 

[Darkbreeze]

Those games are ALL highly threaded, and core/thread dependent. FPS and some other physical factors in those games are all pretty closely tied to core counts with hyperthreads adding some additional benefits as well.

Te XFX XT power supplies are not made by Seasonic like the rest of the XFX products have traditionally been. They are made by Shenzhen Ruishengyuan and they are poor quality. It would not be surprising if you were suffering from some loss of performance due to any of various PSU related factors such as bad voltage regulation or high ripple levels. It may not even be capable of adequately supplying the full required capacity which could definitely translate as a loss of performance on both the CPU and graphics card, especially under very high demand loads.

If you have no exhaust fans, especially in that case or one like it, you can just about guarantee you are overheating the motherboard VRMs. The fact that you have intake fans and no exhaust fans is completely backwards of the way you SHOULD have it configured. Positive pressure is ONLY beneficial if you already have plenty of exhaust airflow. Exhaust airflow is a required component of cooling performance since it doesn't matter how much air you get into the case if you are not getting the heat OUT of the case. If you have to choose between negative and positive pressure, where cooling is a concern, always choose negative or neutral pressure configurations as negative pressure increases cooling performance while positive pressure optimizes dust suppression. A balance is nicer, but if you have to choose, choose exhaust fans over intake fans. It's easy for cool air to find it's way into the case if you are sucking all the pressure out of the case.

The outside air will already be higher pressure and will force it's way in if necessary. Trying to push the heat OUT of the case though, using only intake fans to do it, is fundamentally flawed and will never work well.

Move those fans to the rear and top-rear locations and orient them as exhaust fans OR get two new fans and put them there as exhaust fans in addition to the fans you already have installed as intake. That would be the better option really, by far.

Update you BIOS to the latest version. That alone might make a ton of difference with that RTX card. It has for many others.

What Windows version are you running, 7 or 10?

 

[HeathenHeart]

I'm running Windows 10.

I've updated the chipset drivers and the BIOS but no effect.

I'm trying to monitor the VRM temperature but the sensor seems off, one of those is marking 0° while the other is at -125°.

Would a homemade solution (running the case open with a big external fan ) be good enough for at least testing if it is a refrigeration issue?

Any way of testing the PSU or the VRM for these effects mentioned?

 

[Darkbreeze]

What are you using to monitor the VRM and other temperatures? I recommend using HWinfo.

Monitoring software​

HWmonitor, Open hardware monitor, Realtemp, Speccy, Speedfan, Windows utilities, CPU-Z, NZXT CAM and most of the bundled motherboard utilities are often not the best choice as they are not always accurate. Some are actually grossly inaccurate, especially with certain chipsets or specific sensors that for whatever reason they tend to not like or work well with. I've found HWinfo or CoreTemp to be the MOST accurate with the broadest range of chipsets and sensors. They are also almost religiously kept up to date.

CoreTemp is great for just CPU thermals including core temps or distance to TJmax on older AMD platforms.

HWinfo is great for pretty much EVERYTHING, including CPU thermals, core loads, core temps, package temps, GPU sensors, HDD and SSD sensors, motherboard chipset and VRM sensor, all of it. When starting HWinfo after installation, always check the box next to "sensors only" and de-select the box next to "summary".


Run HWinfo and look at system voltages and other sensor readings.

Monitoring temperatures, core speeds, voltages, clock ratios and other reported sensor data can often help to pick out an issue right off the bat. HWinfo is a good way to get that data and in my experience tends to be more accurate than some of the other utilities available. CPU-Z, GPU-Z and Core Temp all have their uses but HWinfo tends to have it all laid out in a more convenient fashion so you can usually see what one sensor is reporting while looking at another instead of having to flip through various tabs that have specific groupings, plus, it is extremely rare for HWinfo to not report the correct sensor values under the correct sensor listings, or misreport other information. Utilities like HWmonitor, Openhardware monitor and Speccy, tend to COMMONLY misreport sensor data, or not report it at all.

After installation, run the utility and when asked, choose "sensors only". The other window options have some use but in most cases everything you need will be located in the sensors window. If you're taking screenshots to post for troubleshooting, it will most likely require taking three screenshots and scrolling down the sensors window between screenshots in order to capture them all.

It is most helpful if you can take a series of HWinfo screenshots at idle, after a cold boot to the desktop. Open HWinfo and wait for all of the Windows startup processes to complete. Usually about four or five minutes should be plenty. Take screenshots of all the HWinfo sensors.

Next, run something demanding like Prime95 (With AVX and AVX2 disabled) or Heaven benchmark. Take another set of screenshots while either of those is running so we can see what the hardware is doing while under a load.

*Download HWinfo


For temperature monitoring only, I feel Core Temp is the most accurate and also offers a quick visual reference for core speed, load and CPU voltage:


*Download Core Temp



​

Ryzen master for Zen or newer AMD CPUs, or Overdrive for older Pre-Ryzen platforms (AM3/AM3+/FM2/FM2+)

For monitoring on AMD Ryzen and Threadripper platforms including Zen or newer architectures, it is recommended that you use Ryzen master if for no other reason than because any updates or changes to monitoring requirements are more likely to be implemented sooner, and properly, than with other monitoring utilities. Core Temp and HWinfo are still good, with this platform, but when changes to CPU micro code or other BIOS modifications occur, or there are driver or power plan changes, it sometimes takes a while before those get implemented by 3rd party utilities, while Ryzen master, being a direct AMD product, generally gets updated immediately. Since it is also specific to the hardware in question, it can be more accurately and specifically developed without any requirement for inclusion of other architectures which won't be compatible in any case. You wouldn't use a hammer to drive a wood screw in (At least I hope not) and this is very much the same, being the right tool for the job at hand.

As far as the older AMD FX AM3+ platforms including Bulldozer and Piledriver families go, there are only two real options here. You can use Core Temp, but you will need to click on the Options menu, click Settings, click Advanced and put a check mark next to the setting that says "Show Distance to TJmax in temperature fields" and then save settings and exit the options menu system. This may or may not work for every FX platform, so using AMD Overdrive is the specific, again, right tool for the job, and recommended monitoring solution for this architecture. Since these FX platforms use "Thermal margins" rather than an actual "core/package" temp type thermal monitoring implementation, monitoring as you would with older or newer AMD platforms, or any Intel platform, won't work properly.

For more information about this, please visit here for an in depth explanation of AMD thermal margin monitoring.

Understanding AMD thermal margins for Pre-Ryzen processors





*Download Ryzen Master



*Download AMD Overdrive


Also, posting screenshots, when requested, is helpful so WE can see what is going on as well and you can learn how to do that here:

How to post images on Tom's hardware forums

​

 

[mdd1963]

Was this a fresh install and test, or, after replacing some other GPU?

Lots to go wrong with the latter path...

Not that the 4790K is any sort of powerhouse, it's no R3-1200 either. You'd have to check known valid reviews with a given game in question with that CPU that include minimum and average FPS to see if the unsatisfactory framerates might be a result of your CPU....