Question High-end system, poorish performance

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Hey guys,

Playing CS:GO I've been having a lot of issues lately with my FPS dipping way lower than it should I'd think with my specs. The number itself isn't exactly what bothers me, but the impact it's having on the game.

When my FPS dips to around 200 or loss (seen it get as low as 160ish I think), my mouse becomes kinda "floaty" and it looks like my game loses 144hz and you can basically see the "choppiness" which I know doesn't make sense because it's still over 160 FPS...but I'm tellin' you that's the best way to describe it.

I've reformated PC, reverted NVIDIA drivers, done all the in-game and NVIDIA video settings, turned off all the Xbox stuff, Windows is performance mode along with NVIDIA setings, etc.

Specs:
MSI z270 Gaming Pro Carbon
i7 7700k
GTX 1080ti aero
32GB DDR4 2400 RAM
 

Eximo

Titan
Herald
Sounds like stuttering.

Have you been monitoring temperatures while this is happening?

7700k are somewhat notorious for running a little warm. What kind of cooler are you using? If it is an AIO, make sure it is still pumping properly.

I doubt the GPU would be overheating running CS:GO even at 200FPS. But always good to keep an eye on.

Also monitor the 12V+ during a heavy load. If it is dipping below 12 more than a .1 or .2 volts your powersupply might be running too warm, or is in need of replacement.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Sounds like stuttering.

Have you been monitoring temperatures while this is happening?

7700k are somewhat notorious for running a little warm. What kind of cooler are you using? If it is an AIO, make sure it is still pumping properly.

I doubt the GPU would be overheating running CS:GO even at 200FPS. But always good to keep an eye on.

Also monitor the 12V+ during a heavy load. If it is dipping below 12 more than a .1 or .2 volts your powersupply might be running too warm, or is in need of replacement.
Thanks for the response.

Here's a screenshot of temps and what not after running an FPS benchmark in CS:GO (like 3 minutes of action testing FPS).

View: https://imgur.com/a/KkemTm0


Here's a screenshot of HWMonitor/CPU-Z just hanging out on the desktop for a few minutes:

View: https://imgur.com/a/EGEz9bn
 

Eximo

Titan
Herald
Well the CPU information isn't all that valuable. Load temperatures, voltages, etc are what matter. Max of 1.32 volts is a little high on the CPU voltage, but not really out of spec. And typical of some motherboard's stock settings.

12V voltage under load would be interesting to see as well.

Your 3 minute GPU test seems fine and an impressive result, but not really long enough to say much. Cooler didn't get saturated and the GPU didn't look like it throttled any, boosted to a reasonable clock speed. Without a multiplayer element it also doesn't accurately reflect CPU load.

Only other common cause for stuttering is disk I/O getting saturated. Probably not likely with CS:GO though, the maps are relatively small.

You can set MSI Afterburner and HWMonitor to log to file and play normally. With the FPS tracked you could actually observe what is happening to all the hardware when that happens.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well the CPU information isn't all that valuable. Load temperatures, voltages, etc are what matter. Max of 1.32 volts is a little high on the CPU voltage, but not really out of spec. And typical of some motherboard's stock settings.

12V voltage under load would be interesting to see as well.

Your 3 minute GPU test seems fine and an impressive result, but not really long enough to say much. Cooler didn't get saturated and the GPU didn't look like it throttled any, boosted to a reasonable clock speed. Without a multiplayer element it also doesn't accurately reflect CPU load.

Only other common cause for stuttering is disk I/O getting saturated. Probably not likely with CS:GO though, the maps are relatively small.

You can set MSI Afterburner and HWMonitor to log to file and play normally. With the FPS tracked you could actually observe what is happening to all the hardware when that happens.
Thanks again!

Okay, so I'm downloading MSI Afterburner. Should I try anything out with it first or just set it to "log to file" and play a game?
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well the CPU information isn't all that valuable. Load temperatures, voltages, etc are what matter. Max of 1.32 volts is a little high on the CPU voltage, but not really out of spec. And typical of some motherboard's stock settings.

12V voltage under load would be interesting to see as well.

Your 3 minute GPU test seems fine and an impressive result, but not really long enough to say much. Cooler didn't get saturated and the GPU didn't look like it throttled any, boosted to a reasonable clock speed. Without a multiplayer element it also doesn't accurately reflect CPU load.

Only other common cause for stuttering is disk I/O getting saturated. Probably not likely with CS:GO though, the maps are relatively small.

You can set MSI Afterburner and HWMonitor to log to file and play normally. With the FPS tracked you could actually observe what is happening to all the hardware when that happens.

So here's the graphs when I view the log:

View: https://imgur.com/a/UUyx1jn
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well the CPU information isn't all that valuable. Load temperatures, voltages, etc are what matter. Max of 1.32 volts is a little high on the CPU voltage, but not really out of spec. And typical of some motherboard's stock settings.

12V voltage under load would be interesting to see as well.

Your 3 minute GPU test seems fine and an impressive result, but not really long enough to say much. Cooler didn't get saturated and the GPU didn't look like it throttled any, boosted to a reasonable clock speed. Without a multiplayer element it also doesn't accurately reflect CPU load.

Only other common cause for stuttering is disk I/O getting saturated. Probably not likely with CS:GO though, the maps are relatively small.

You can set MSI Afterburner and HWMonitor to log to file and play normally. With the FPS tracked you could actually observe what is happening to all the hardware when that happens.
View: https://imgur.com/a/6ICMTVI


That's the full MSI afterburner log.

Anyone?
 

Eximo

Titan
Herald
CPU is certainly getting worked hard, but that is to be expected at high frame rates. Doesn't look like it is overheating, power output is actually fairly low for the voltage.

Did you experience the problem during these sections of the log?
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
CPU is certainly getting worked hard, but that is to be expected at high frame rates. Doesn't look like it is overheating, power output is actually fairly low for the voltage.

Did you experience the problem during these sections of the log?
Thanks again for the response.

Yes, I did.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
CPU is certainly getting worked hard, but that is to be expected at high frame rates. Doesn't look like it is overheating, power output is actually fairly low for the voltage.

Did you experience the problem during these sections of the log?
Tried using the "game boost" option in BIOS and it put me up to 4.8ghz but my FPS was worse..so...yeah
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well the CPU information isn't all that valuable. Load temperatures, voltages, etc are what matter. Max of 1.32 volts is a little high on the CPU voltage, but not really out of spec. And typical of some motherboard's stock settings.

12V voltage under load would be interesting to see as well.

Your 3 minute GPU test seems fine and an impressive result, but not really long enough to say much. Cooler didn't get saturated and the GPU didn't look like it throttled any, boosted to a reasonable clock speed. Without a multiplayer element it also doesn't accurately reflect CPU load.

Only other common cause for stuttering is disk I/O getting saturated. Probably not likely with CS:GO though, the maps are relatively small.

You can set MSI Afterburner and HWMonitor to log to file and play normally. With the FPS tracked you could actually observe what is happening to all the hardware when that happens.
These stats on voltages/temps are after running a CPU stress test via the Intel Extreme tuning utility.

View: https://imgur.com/a/1aVEuS9
 

Eximo

Titan
Herald
Well, a 2400Mhz memory kit is certainly below par for your typical unlocked CPU. What is the latency on it? If it is above 15 that is pretty poor for a 2400Mhz kit. Very easy to get 3000Mhz kits with a latency of 15, 3200 is usually 16. I think there are a few Kingston kits out there at 2400Mhz with a latency of like 12 or 13, so if they made it into userbenchmark's database that is what you are getting compared to.

Keep in mind that most of the people that bother to run this benchmark with K CPUs are going to have overclocked. Probably why your CPU and GPU aren't getting top marks, just near the top.

Doesn't look like you have any power sag at all, pretty uncommon to see EXACTLY 12.00 and 5.00 volts. Must be a well regulated power supply.

About the only thing I can think of is the fully populated memory. Try taking out two sticks and see if it does any better.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well, a 2400Mhz memory kit is certainly below par for your typical unlocked CPU. What is the latency on it? If it is above 15 that is pretty poor for a 2400Mhz kit. Very easy to get 3000Mhz kits with a latency of 15, 3200 is usually 16. I think there are a few Kingston kits out there at 2400Mhz with a latency of like 12 or 13, so if they made it into userbenchmark's database that is what you are getting compared to.

Keep in mind that most of the people that bother to run this benchmark with K CPUs are going to have overclocked. Probably why your CPU and GPU aren't getting top marks, just near the top.

Doesn't look like you have any power sag at all, pretty uncommon to see EXACTLY 12.00 and 5.00 volts. Must be a well regulated power supply.

About the only thing I can think of is the fully populated memory. Try taking out two sticks and see if it does any better.
UserBenchmark says:
Latency 65.5

61% 65.5 ns
View: https://imgur.com/a/cIdajb4


This is the RAM I have:
AVANT 8GB DDR4-2400 AVW641GU48J7240ND-SPXA6

However, it still says overall it's performing above expectations (73rd percentile).

I actually tried the RAM thing last night. I took out one stick (slot1)...didn't change..then took out slot2 as well..nothing. So I put them back. I can try slots 3 and 4 today if you think I should.

I ruled out the GPU...
CPU temps/speeds look fine...
RAM clocks at above expectations...

:( :( dropping to 140fps in CS:GO with my system makes no sense lol
 

Eximo

Titan
Herald
Should be something like slot 1 and 3 or 2 and 4 or A1, B1, A2, B2. Depends on how they are labeled. But the ram sticks should always be separated by a gap to be in dual channel mode. Typical config should be the slot furthest from the CPU, and one slot empty, and then the next slot filled.

I can't find a spec sheet for that memory, I can only seem to find it on ebay and the manufacturer has no lists of their products.

It might be unbuffered ECC memory, which isn't technically supported?

Try and run CPU-Z and look at the memory section. 65.5ns seems way too slow (I used to buy 70ns EDO memory back in the mid-90s)

You might want to run back to best buy and pick up a 2x8GB memory kit and see how that goes as well.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Should be something like slot 1 and 3 or 2 and 4 or A1, B1, A2, B2. Depends on how they are labeled. But the ram sticks should always be separated by a gap to be in dual channel mode. Typical config should be the slot furthest from the CPU, and one slot empty, and then the next slot filled.

I can't find a spec sheet for that memory, I can only seem to find it on ebay and the manufacturer has no lists of their products.

It might be unbuffered ECC memory, which isn't technically supported?

Try and run CPU-Z and look at the memory section. 65.5ns seems way too slow (I used to buy 70ns EDO memory back in the mid-90s)

You might want to run back to best buy and pick up a 2x8GB memory kit and see how that goes as well.
Yeah, I meant I did 1/3 out, not 1/2, so that would have made it exactly as you described in terms of the slot furthest from CPU filled and the second closest filled.

It COULD be that the latency shown on the userbenchmark is the clock speed (16.0 clocks in CPU-Z) x4 (4 sticks of RAM).

16 CL x 4 = 64.

This is CPU-Z of SPD/Memory from a couple days ago:

View: https://imgur.com/a/n5pMM9r
 

Eximo

Titan
Herald
Well, either of those settings, 17 or 18 for CAS latency, pretty poor for 2400Mhz memory.

15-17-17-35 would be a good kit of 3000Mhz.

Yours is

17-17-17-39 or 18-17-17-39 at 2400Mhz

You are running 16-18-18-35 at 2666Mhz possibly?

If you consult this chart you should have a true latency around 14ns with a CAS ot 17 at 2400Mhz, so 65.5 ns points to something being very off.

https://en.wikipedia.org/wiki/CAS_latency

No idea how accurate userbenchmark is in that regard though. Might be something in the BIOS causing it to be slow, or the memory controller on the CPU has some sort of issue.
 

steventhehro

Honorable
Aug 21, 2012
81
0
10,630
0
Well, either of those settings, 17 or 18 for CAS latency, pretty poor for 2400Mhz memory.

15-17-17-35 would be a good kit of 3000Mhz.

Yours is

17-17-17-39 or 18-17-17-39 at 2400Mhz

You are running 16-18-18-35 at 2666Mhz possibly?

If you consult this chart you should have a true latency around 14ns with a CAS ot 17 at 2400Mhz, so 65.5 ns points to something being very off.

https://en.wikipedia.org/wiki/CAS_latency

No idea how accurate userbenchmark is in that regard though. Might be something in the BIOS causing it to be slow, or the memory controller on the CPU has some sort of issue.
Alright, so when I returned the GPU I bought to test it I went ahead and picked up some RAM. I bought Corsair Vengeance 2x8 DDR4-3200. Put it to XMP in BIOS.

There were improvements. My FPS is better but I'm still hitting some crazy lows. Like 1 big swings from 300-350fps down to 150, mouse starts to drag/get sluggish as the fps is low, screen is like rippling/stuttering. Then it comes back. This is typically in situations with molotovs thrown, smoke grenades thrown, several players, etc.
 

Eximo

Titan
Herald
That doesn't sound too unreasonable. I've seen footage of 400FPS while staring at a floor, and it changing rapidly down as players and more of the map is in view. The sluggishness could just be your perception of the frame rate changing so much.

You might try posting in some CS:GO centric forums to see about advice on game settings.
 

ASK THE COMMUNITY

TRENDING THREADS