Low FPS gaming (CS:GO) with FX8350 and R9 290x

Toggle sidebar Toggle sidebar
A

alansupra94

Prominent
Jun 16, 2017
15
0
520
Well after doing extensive research, I have noticed that I am not getting close to the FPS I should be getting while gaming in CS:GO.

Computer specs:
Asus ROG Crosshair Formula V
16GB of DDR3 G-skill 2100mhz
AMD 8350
R9 290 (flashed to 290x)
3 SSD and 4TB
750W powersupply
144hz Asus monitor

No matter what setting I use (graphics wise) I am getting 60-100fps with dropping to 40fps randomly. What should I check first? I also may check to see if there is a bios update as well as make sure I flashed the proper bios is on the GPU.
 
Solution
A
Well I wanted to update everyone on what I discovered:

So after getting fed up, a friend was find enough to lend me his computer which was very similar to mine. I found he was getting the same benchmarks in cs:go and the reason was I was using the NEW maps which are not fully optimized. So my system was fine at stock levels. It didn't explain why I was struggling to overclock.

I ended up deciding that maybe I should put new thermal paste onto the cpu. I took the xigametek cooler off and noticed that the thermal paste was kinda squeezed out. I decided to investigate and found out that the cooler was sitting about 1mm-2mm too close to the CPU that there was very very very little if any thermal paste between the cooler and cpu. The cpu...
Sort by date Sort by votes
Cs:go doesn't use a lot of cores, generally it relies on a couple of fast cores. The 8350 suffers from lower ipc so despite having 8 cores (4 modules) it might lose out to something like an i3 with just a couple faster cores. Have you checked temps? Try using amd overdrive and see what the thermal margin is, it reads out how many degrees until it reaches throttle (overheating).
 
Upvote 0 Downvote
It may not be overheating but if it is for some reason it could be thermal throttling. When a cpu thermal throttles it automatically reduces clock speed to counter higher temps which can result in fps drops. You may also want to check your power plan settings in windows and see if it's set to performance mode to make sure the cpu is getting full power all the time.
 
Upvote 0 Downvote
That seems fine on temps, the margin at 26.6c suggests you've got that many degrees to go until you're thermal throttling and the frequency meter show's you're reaching 4.2ghz. You could try overclocking the cpu a bit and see if that helps but none of the cores look maxed out. Try running msi afterburner while gaming, see what it says during gaming with the on screen display (osd). See if the gpu is running at 100%. I'd be surprised if it was but worth a look.

Did you make sure you're running on performance mode under the power plan settings? Afterburner should also help you monitor your gpu temps. It should be lower than 90c preferably, if it's hitting 90-95c it may be the gpu thermal throttling.
 
Upvote 0 Downvote


Any ideas? The GPU is properly cooling. I tried overclocking both the CPU and GPU and saw no change. One thing I did notice is why is the memory clock on the memory 500mhz? Shouldn't it be more?
 
Upvote 0 Downvote
Your temps sound great, leading me to suspect either the PSU or drive integrity. The latter is where I'd start.

Defrag your HDD's, trim the SSD's. Window's drive optimization tool is good enough. Just select each drive (well, volume if you're raiding anything) and click the optimize button. Or you can schedule a weekly optimization or something like that. I just do them manually whenever I remember to.

Download Memtest86 and run it. Or you can try out window's memory diagnostic tool before running memtest86. This'll tell you if your memory has gone awol or shot itself in the foot or something.

A PSU that is dying, but hasn't catastrophically failed and taken your entire system with it, will have symptoms similar to this. And so is a CPU on the cusp of not getting enough voltage, but not so little it's actually unstable. Or a CPU getting an unstable voltage. It's like a car running on fumes, coughing every time you go over a small bump in the road. The bumpier the road, the less empty the fuel tank needs to be for the coughing to happen.

4.2ghz stable at 1.35V~ish seems pretty reasonable to me. You have to take Vdroop into account when the CPU is under full load, made worse by the bulldozer/piledriver architecture's power-hungry nature and inefficient use of said power. I see in one screenshot that you've assigned the CPU 1.41V, but in the Overdrive screenshot it's only running 1.33? That is some oddly excessive Vdroop for a board of that quality. Either that, or your board reads Vcore and VID in overdrive differently from my late m5a99x.

DDR stands for "Double Data Rate", so that 543 mhz is the base frequency. You x2 it for the effective frequency, which in your case is about 1086, where 1066 is the standard operating speed for all DDR3, so, close enough, I guess? You'll have to manually change the ram frequency value or enable XMP to get the advertised ram speeds.

Speaking of close enough, I'm curious as to why your reference clock (base clock) isn't at the standard 200. Did you try to use the auto OC utility? That old m5a99x had a standard reference clock was 200, even though I ran it near 300, so that 203.69 bugs me a little if you didn't manually change that value.

Make sure AMD Cool'n'Quiet and spread spectrum and all those power-saving/legality features are disabled in BIOS. I can't remember all of them off the top of my head, haven't played with FX CPU's in a few years. In fact, you may want to lock the CPU multiplier at your overclock value just for testing. You can make it ramp up and down as needed later.
 
Upvote 0 Downvote
Thank you for your response. That screenshot is from when I let the built in Asus
"Level-up" overclock my PC so whatever it selected, I let it run. It literally only went up 1-2 levels before losing stability. I read about people getting 4.8ghz on air and I can't get 4.3ghz. I put on new thermal paste. I did notice that when I overclocked, the delta temperature was fluctuating rather large (from 5C to 15C) while gaming in CS:GO.

I am going to run memory test (as I have another set of G.Skill DDR3 1600) and I will post the screenshot. I will see if I have time to drop into microcenter and maybe pick up a new powersupply or try using the 1500W one I had in before and swapped out.

Thanks for responding to me!

P.S. My last ditch option is going to be buying a used motherboard and cpu and just either selling or throwing this setup into the trash lol.
 
Upvote 0 Downvote
That explains the really weird numbers I'm seeing. Go into bios, reset everything to factory. Auto OC utilities are pretty horrible. I would never trust them with any of my builds.

You could switch to Ryzen (R5 1600 gets my vote) to absolve yourself of this miserable steaming heap of sh t called the FX CPU, but CS:GO at 1080p should not be running under 100fps with an R9 290x. I mean, I was well over 150fps most of the time when I attempted to run it on an Xeon W3540 (I7 940 with iGPU disabled) at 1440p with 12GB of 1033mhz ECC ram (triple channel, really old stuff) and an HD7950. On a Mac. With Bootcamp. Stuck at PCIe Gen 1 speeds and no overclocking whatsoever because, well, Apple. All of which is inferior in just about every way to your rig.

I still highly suspect drive integrity. Please defrag your HDD's and optimise your SSD's and see if gaming performance doesn't improve.

Oh, and power supply wattage is not an indication of power supply quality. I'd much rather run a system on a high quality 450W power supply than a questionable 1500W unit. For futher reading, check out this here link.

If you are going to try and overclock, do it manually, the good old fashioned way. Well, actually, you can sorta cheat if you've got a good CPU cooler (I still don't know what cooler you're using), just pop the core voltage to 1.4 and the core multiplier at 21, reboot, and test for stability (AIDA64, Prime95, OCCT, whatever you want) while monitoring temps with AMD Overdrive. If it runs for 30 minutes + without crashing or hiccuping or BSOD'ing or overheating, then maintain core voltage and up the multiplier by 1, and test for stability again. Keep upping the core multiplier until it is unstable, then up the voltage by 0.01, and keep going until you either hit 1.5V or start overheating. I wouldn't recommend you go playing around with reference clock overclocking just yet 😉 . And while you're at it, turn off every power-saving feature you can find in bios (if you don't know what a setting is, just google it or leave it alone). Turning off cool'n'quiet and spread spectrum alone is enough to greatly increase CPU overclock stability. I vaguely remember ASPM or something, but I think that's PCIe slot-related power fluctuation. Oh, and keep XMP disabled and all RAM settings at stock while CPU overclocking. You can turn it back on later. Tends to mess with CPU stability when you're pushing the SB really hard.

Just, don't do what I did and attempt to daily an FX octacore at 5.1ghz@1.62~ish Vcore. My poor watercooled motherboard commit suicide in a rather explosive way. I even disabled every other core per module to hopefully push core and cache speeds higher. 5.1ghz core and 4.6ghz NB was the fastest I could get. Then again, that was the older FX8320. I've never tried on the '50 series.
 
Upvote 0 Downvote
Thanks so much for responding guys.

The 1500W is a industrial grade model with a 20 year warranty. I might swap it today.

I will check the ram (I will run memtest98) and see what it says. I will check the SSD for integrity. How do I optimize a SSD?

Honestly if I can get 200fps in all maps on CS:GO I won't bother trying to overclock haha.

I will report back and tell you guys what I find.
 
Upvote 0 Downvote
This was buried in my earlier wall of text. I haven't found any good third party tools for optimizing SSD's though.

Oh, and while you're at it, download CCleaner and run both the normal cleaner and the registry cleaner.

 
Upvote 0 Downvote


Awesome thank you I will try that. I will run CCleaner too. My ram is DDR3 1866 Gkills Sniper. Do you think I would be better off putting in DDR3 1600 Gkills (I have that in a box)?
 
Upvote 0 Downvote


Okay I will. I really hope this helps. I may end up re-installing Windows 8.1 if it doesn't help.
 
Upvote 0 Downvote
So memory check came back fine. I notice that my VID and Frequency kinda of move around by itself. I also notice that my CPU in CS:GO it goes to 50-70% usage. It also seems to be using core 5.

iwt5k9.jpg
 
Upvote 0 Downvote
So update:

This PC has been pissing me off royally to the point where I think I might just sell it.

I have done the following:

- Installed Windows 10 (no change)
- Installed Windows 10 on a different SSD (no change)
- Installed CS:GO fresh (no change)
- Installed CS:GO fresh on a new SSD (no change)
- Removed ram and reseated (no change)
- Installed 1500W powersupply (no change)
- Dusted case (no change)

I have narrowed everything down and its to the point where it is either the motherboard/CPU/GPU and rather than trying to solve it, I think I am just going to buy component from Microcenter and throw this off a cliff. If you guys have any more ideas, I would love to hear it.

Last things I may do is:

1. Remove any cs:go configs
2. Try a new processor and graphics card (I think I have a spare GPU somewhere)
 
Upvote 0 Downvote
Check that your motherboard slow mode/LN2 switch isn't in the on position.
Clear CMOS.
Did you accidentally enable some kind of in-game frame limiter?
Is your motherboard BIOS up to date?
graphics drivers up to date?
Are the useless windows services fully disabled, like Superfetch, Xboxlive anything, windows notifications, etc.?

I had a wonky gigabit ethernet card driver murder the crap out of my old FX system. Every time I rebooted, it'd register itself as a new network device, so my computer was eventually left with hundreds of network devices it was constantly scanning for, bringing the whole thing to a grinding halt. In the end, I had to delete all several hundred entries manually, which took an entire afternoon, but that brought me my system back.

The last thing that comes to mind before I'm completely dry on ideas is AMD Powertune's notoriety for downclocking cards via bios updates to hit a certain power target. The R9 380x/390/390x immediately comes to mind, where my Sapphire Nitro 380x model refused to ever clock over 900mhz. Its stock clock was supposed to be 1040mhz, it never went over 65C at any time. I even tried to manually overclock, mashed the power target and core clock sliders all the way to the right (150% power target, 1200mhz core. And still it stupidly sat at 900mhz at 58C. Had to use ClockBlocker to force the card to sit at 100% speed/load the whole time for it to ever actually reach full performance. AMD never fixed this thing since it was introduced in like 2010. And it has murdered the performance of many generations of graphics cards, from the HD6XXX series all the way to the R9 390x/Fury/Fury X. All because they wanted to match Nvidia in "efficiency". Thankfully anything that's Polaris and newer doesn't have this exact problem. It has a different, but similar problem, thanks to invisible settings in Wattman. The worst part is that this doesn't affect every card. My HD7950 had no Powertune issues, but my 380x was plagued by it. You lose the lottery, you get slapped around by Powertune until you permanently switch to Nvidia. This is what turned me off from buying AMD graphics cards for good.

okay enough ranting.
 
Upvote 0 Downvote


1. I will check the motherboard slowmode/Ln2 switch. Where is that located?
2. I don't think I enabled anything in cs:go
3. Bios is the most to date revision.
5. I have disabled everything.
6. I don't think the CPU is getting taxed because the usage is only going to 50-60% max.

I will check powertune. I know that this is a reference card R9 290 that was unlocked and bios flashed to a R9 290x. I will check the bios again on it but I believe it is perfectly fine.


 
Upvote 0 Downvote


I don't know where that switch is on that particular board. Consult the motherboard manual. Or Google.

50-60% is pretty near maxed for an FX when gaming. It's the way the architecture works. An FX 8-core CPU is actually a quad-core CPU, but instead of hyperthreading like Intel, they stuck a physical extra core in there instead. A horrible idea, horribly implemented, leading to the introduction of the piledriver module-based architecture (2 cores tied to a single cache per module). And then they made it worse by introducing Bulldozer, a.k.a. a hotter piledriver. It got so hot that the solder between the IHS and CPU die became the biggest thermal bottleneck. And then the godawful FX 9XXX series came out. It was an absolute abomination, and shipped from the factory with an AIO. Games are typically (no, always) coded around Intel CPU architecture, which assumes a single strong core with which to load with threads and calculations, and hyperthreading for parallelization. AMD's piledriver had no strong cores, and was designed to have two weak cores run really fast and constantly alternate back and forth. This is the 50-60% you're seeing, since the program is only trying to use every other core. This is why everyone says the FX CPU's were horrible gaming CPU's, and which is also why Intel was the only real option for gaming for nearly 10 years until Ryzen came along. Even Vulkan/DX12, touted to fix the design failure of not utilizing more than half of the cores in those FX CPU's, didn't work. Rather, it backfired. Thus, most games run worse in DX12 mode than DX11. And Vulkan only has like 3 major titles.

Powertune isn't in the GPU bios. It's a specific chip soldered to the voltage controllers. If you take a soldering iron and melt the chip off or short it out with more solder, you can disable it. The GPU Bios basically just takes readings off this chip to throttle itself. The easiest way to find out if you're downclocking is to download and run Clockblocker, and see if your performance changes.

Anyway, I'm all dry on ideas now. Other than a corrupted ISO of Windows 10.
 
Upvote 0 Downvote
Well I wanted to update everyone on what I discovered:

So after getting fed up, a friend was find enough to lend me his computer which was very similar to mine. I found he was getting the same benchmarks in cs:go and the reason was I was using the NEW maps which are not fully optimized. So my system was fine at stock levels. It didn't explain why I was struggling to overclock.

I ended up deciding that maybe I should put new thermal paste onto the cpu. I took the xigametek cooler off and noticed that the thermal paste was kinda squeezed out. I decided to investigate and found out that the cooler was sitting about 1mm-2mm too close to the CPU that there was very very very little if any thermal paste between the cooler and cpu. The cpu was getting huge flucuations, becoming unstable and shutting down.

I went down to my garage, installed 4 washers (one on each post) and it raised it about 1.5mm. Paste is now evenly on the surface of both the GPU and cooler. I also decided to just re-do all the thermal paste on the VRMs and chips with Artic Silver. I also threw out my Antec P182 case since it was so dusty and seemed to have not the best airflow. I put everything into another Antec case that has a ton of fans including a 200mm one right over the heatsink of the CPU. I also added another 120mm fan on the other side of the CPU cooler for added flow.

Now my PC is at 4.8ghz and 1.45V without a hiccup. I gained about 30fps in CS:GO on optimized maps. Couldn't be happier. Thank you guys for all the help.
 
Upvote 0 Downvote
Solution
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom