e5-2667v2 bottleneck?

[artavasdes]

Hello, I have a computer with dual e5-2667v2 watercooled clocked to 3.7 ghz. They are on a supermicro x9dri-f motherboard, 128 gb 1866 mhz ECC ram and a Radeon VII. In games like Rainbow Six Siege, I get around 150 FPS on maxed out settings @ 1080p, according to other benchmarks, I am supposed to be getting around 250+ FPS. In Division 2, I get 60 FPS at maxed settings @ 1080p (I am supposed to be getting around 110-120). In Ghost Recon Breakpoint, I get around 60 FPS at maxed out settings @ 1080p while I am supposed to get around 110 FPS. I thought this was a bottleneck, but after watching this video (

View: https://www.youtube.com/watch?v=iz8ThsFOP3w
) , I think it might be something else because the e5-2667v2 is a slightly better cpu than the e5-2689 (which was used in the video). Can it be a GPU defect? Thanks!
Also, here is a spec comparison between the e5-2667v2 and the e5-2689.

 

[Eigenvektor]

60 fps sounds like it might be v-sync, if you have a 60 Hz monitor. Did you check whether that was disabled? Try to monitor GPU and CPU usage (individual cores) while gaming, that should tell you if you run into a bottleneck.

[artavasdes]

@Eigenvektor I have a 240 hertz monitor and it sucks to see none of these games be able to run at 240 FPS, lol. I will try to do the GPU and CPU usage on individual cores, while doing that, what are somethings that will show a cpu bottleneck?

Thanks!

[Eigenvektor]

Just now, artavasdes said:

@Eigenvektor I have a 240 hertz monitor and it sucks to see none of these games be able to run at 240 FPS, lol. I will try to do the GPU and CPU usage on individual cores, while doing that, what are somethings that will show a cpu bottleneck?

Thanks!

If you have a CPU core (or multiple cores) running at 100% and your GPU is not running at 100% that suggests a bottleneck. You can use something like MSI Afterburner's OSD to look at usage while gaming.

[artavasdes]

Thank you, when I get on my computer later this day, I will make sure to try it out and tell you what I get.

[artavasdes]

@EigenvektorI ran the Rainbow Six siege benchmark and my cpu was at 45% load while my gpu was at 100% load. Is it possible that the Radeon VII defected?

[Eigenvektor]

9 hours ago, artavasdes said:

@EigenvektorI ran the Rainbow Six siege benchmark and my cpu was at 45% load while my gpu was at 100% load. Is it possible that the Radeon VII defected?

Could be, but a defect that simply lowers performance seems unlikely, with the possible exception of cooling.

 

How are the temps and clock rate of the card during the benchmark? Is the card also under water, or air cooled? If it doesn't get enough fresh air or the cooler isn't mounted properly it might be thermal throttling. Might be too little air flow in the case with the CPU under water.

 

You could also use GPU-Z to try and see if anything seems out of the ordinary.

[artavasdes]

@Eigenvektor The case is a thermaltake view 71. Although I am open to cooling being issue, I find it hard to believe that it is taking a 50% toll on my performance. Anyways, the dual e5-2667v2's are water cooled by a 360 60mm radiator for both of them. I have them in a push pull configuration. I don't have the radeon vii overclocked. I can use NZXT cam to find out the temps and other clocks. I have some thermal sensors in the case that are connected to my commander pro, but the iCUE software for it does not open up.

[Eigenvektor]

Just now, artavasdes said:

@Eigenvektor The case is a thermaltake view 71. Although I am open to cooling being issue, I find it hard to believe that it is taking a 50% toll on my performance. Anyways, the dual e5-2667v2's are water cooled by a 360 60mm radiator for both of them. I have them in a push pull configuration. I don't have the radeon vii overclocked. I can use NZXT cam to find out the temps and other clocks. I have some thermal sensors in the case that are connected to my commander pro, but the iCUE software for it does not open up.

Afterburner should also report GPU temps (should be able to see them while gaming, same as fps and usage). If the GPU is thermal throttling you should be able to see high temps there.

 

And, as I said, you could use GPU-Z to see if something looks off.

[artavasdes]

@EigenvektorI played a game for an hour and the gpu did just fine, no thermal throttling. The radeon vii stayed around 65-73 celsius. Also, somebody in another forum is telling me to try with just one cpu, should I do this?

[Eigenvektor]

6 hours ago, artavasdes said:

@EigenvektorI played a game for an hour and the gpu did just fine, no thermal throttling. The radeon vii stayed around 65-73 celsius. Also, somebody in another forum is telling me to try with just one cpu, should I do this?

Hm, I was also wondering if dual-CPU might be an issue. It's not very common in gaming rigs, and maybe that has something to do with it. I know that e.g. Threadripper 2 had some issues in games, due to multiple NUMA nodes and/or games being unable to handle this many cores. It has a "game mode" that disables half the cores. So I suppose it's definitely worth a try.

 

I've also come across another thread were someone had a similar issue and people suggested that it might be a PSU that was too weak, i.e. the card couldn't run at full speed because it couldn't get enough power. Might be something else to look at.

[artavasdes]

8 hours ago, Eigenvektor said:

Hm, I was also wondering if dual-CPU might be an issue. It's not very common in gaming rigs, and maybe that has something to do with it. I know that e.g. Threadripper 2 had some issues in games, due to multiple NUMA nodes and/or games being unable to handle this many cores. It has a "game mode" that disables half the cores. So I suppose it's definitely worth a try.

 

I've also come across another thread were someone had a similar issue and people suggested that it might be a PSU that was too weak, i.e. the card couldn't run at full speed because it couldn't get enough power. Might be something else to look at.

I just wanted to say that I really appreciate the help. I have a pretty good PSU, it is a ROG Thor Platinum rated 1200 watt power supply. It has an OLED wattage reader on the side so you can how much power it is using. It seems odd, but in games like Ghost Recon Breakpoint where I get 60 FPS, the system is only using 350 watts. In games like Rainbow Six Siege, the system is using 500 watts. It will be a pain in the butt to physically pull out one cpu and try it like that, is it possible that I will get the same affect by disabling half the cpus cores and thread via BIOS or is there anyway I can do it. I knew about the issues with threadripper 2 and the different modes it has, is it possible to apply the NUMA thing done with it to my two xeons?

[artavasdes]

While I was in my BIOS trying to look for the option to disable cpu cores, I found out I had hyperthreading enabled, so I disabled it. I ran the rainbow six siege benchmark and instead of getting an average FPS of 155, I got 190. This is a great improvement, but I am still going to try and remove some ram modules (see if it does anything), see if I can disable cpu cores, and remove one cpu. Thanks for the help.
The hyper threading really takes a toll on your performance with cpus like this.

[Eigenvektor]

15 hours ago, artavasdes said:

While I was in my BIOS trying to look for the option to disable cpu cores, I found out I had hyperthreading enabled, so I disabled it. I ran the rainbow six siege benchmark and instead of getting an average FPS of 155, I got 190. This is a great improvement, but I am still going to try and remove some ram modules (see if it does anything), see if I can disable cpu cores, and remove one cpu. Thanks for the help.
The hyper threading really takes a toll on your performance with cpus like this.

You're welcome I suspected you had a good PSU, just thought I'd mention it, because I saw someone else have an issue with it. Good to see disabling HT did improve things a little. Not sure you'll be able to disable a whole CPU in BIOS (don't think its possible), but some have the option to limit the number of cores at least. Seems like a pain, but I suspect you'll have to remove one CPU to see if it improves things further.

 

Not much you can do about NUMA. It's basically how a CPU is attached to its memory. If a core needs to access RAM that physically belongs to the other CPU, it'll have to go through that other CPU, which adds additional latency.

 

 

[artavasdes]

On 12/1/2019 at 2:54 AM, Eigenvektor said:

You're welcome I suspected you had a good PSU, just thought I'd mention it, because I saw someone else have an issue with it. Good to see disabling HT did improve things a little. Not sure you'll be able to disable a whole CPU in BIOS (don't think its possible), but some have the option to limit the number of cores at least. Seems like a pain, but I suspect you'll have to remove one CPU to see if it improves things further.

 

Not much you can do about NUMA. It's basically how a CPU is attached to its memory. If a core needs to access RAM that physically belongs to the other CPU, it'll have to go through that other CPU, which adds additional latency.

 

 

After gaining performance disabling HT, it shows how it is probably because there are too many cores. I just bought a cooler for one cpu when I test it with one cpu. But it is not practical to remove that one cpu from the system every time I am about to game. I tried to disable cores with windows, but it did not work. I am currently waiting for a cooler so I can use it to test the one cpu.

[Eigenvektor]

38 minutes ago, artavasdes said:

After gaining performance disabling HT, it shows how it is probably because there are too many cores. I just bought a cooler for one cpu when I test it with one cpu. But it is not practical to remove that one cpu from the system every time I am about to game. I tried to disable cores with windows, but it did not work. I am currently waiting for a cooler so I can use it to test the one cpu.

True, that seems very impractical. But you just reminded me of something that I didn't think of before: Windows has the ability to limit a program to certain cores.

 

So instead of disabling HT and/or cores, you could try to limit the cores the game is allowed to run on, which should have exactly the same effect. Assign specific processor cores for certain apps in Windows 10

[artavasdes]

8 minutes ago, Eigenvektor said:

True, that seems very impractical. But you just reminded me of something that I didn't think of before: Windows has the ability to limit a program to certain cores.

 

So instead of disabling HT and/or cores, you could try to limit the cores the game is allowed to run on, which should have exactly the same effect. Assign specific processor cores for certain apps in Windows 10

Wow, thank you. This seems great if it works, I will definitely try it once I get on my computer.

[artavasdes]

@Eigenvektor I tried running some games with different cores enabled, I got worse FPS with less cores. I got the highest FPS when I enabled all cores.

[Eigenvektor]

1 hour ago, artavasdes said:

@Eigenvektor I tried running some games with different cores enabled, I got worse FPS with less cores. I got the highest FPS when I enabled all cores.

Hm, darn. Not sure what else could be holding you back I know ECC memory can have slightly worse timing, but that shouldn't hit you this hard, if at all.

[artavasdes]

22 hours ago, Eigenvektor said:

Hm, darn. Not sure what else could be holding you back I know ECC memory can have slightly worse timing, but that shouldn't hit you this hard, if at all.

Yeah, with ecc memory, the difference should just be a couple frames. I am going to test the Radeon VII on another system. Do you think taking one of the cpus out will do anything even though I limited the cores it could run on so theoretically it shouldn't make a difference.

[Eigenvektor]

Just now, artavasdes said:

Yeah, with ecc memory, the difference should just be a couple frames. I am going to test the Radeon VII on another system. Do you think taking one of the cpus out will do anything even though I limited the cores it could run on so theoretically it shouldn't make a difference.

Yeah, I'd also go for testing the GPU in another system first. Or possibly swapping it for a different one if e.g. a friend is willing to borrow you theirs.

 

I'm not sure taking the CPU out is still worth it, if limiting cores did nothing. I suppose it's not quite the same, but close enough? I've also thought about limiting RAM/removing some sticks just to see what it does, but idk if I'm really expecting that to do anything

[artavasdes]

36 minutes ago, Eigenvektor said:

Yeah, I'd also go for testing the GPU in another system first. Or possibly swapping it for a different one if e.g. a friend is willing to borrow you theirs.

 

I'm not sure taking the CPU out is still worth it, if limiting cores did nothing. I suppose it's not quite the same, but close enough? I've also thought about limiting RAM/removing some sticks just to see what it does, but idk if I'm really expecting that to do anything

I can remove some ram, at this point I guess it is worth it to see. I have a 1060 6gb, I will try with that. Thanks.

[artavasdes]

On 12/8/2019 at 10:14 AM, Eigenvektor said:

Yeah, I'd also go for testing the GPU in another system first. Or possibly swapping it for a different one if e.g. a friend is willing to borrow you theirs.

 

I'm not sure taking the CPU out is still worth it, if limiting cores did nothing. I suppose it's not quite the same, but close enough? I've also thought about limiting RAM/removing some sticks just to see what it does, but idk if I'm really expecting that to do anything

I really don't have much time during the weekday to do anything troubleshooting wise, but I definitely will test out some of your recommendations on the weekend I will immediately tell you my results. I'll swap the gpu for another (a 1080) and see if I get bad performance or loss a noticeable amount of FPS. I think this is the best way to see if it is a gpu problem, maybe I'll try the radeon vii on another system. Also, if I the performance that I am getting is bad, I will try removing some of the ram, otherwise, I don't know what it might be.

[artavasdes]

@Eigenvektor I haven't had much time during the weekend to test anything out. This weekend two things I will try include swapping the gpu and removing some ram. I just wanted to ask whether the pcie slot I have the gpu plugged into can make a difference?

[Eigenvektor]

12 hours ago, artavasdes said:

@Eigenvektor I just wanted to ask whether the pcie slot I have the gpu plugged into can make a difference?

It does, but minimal at best. It depends on how many PCIe lanes the slot has, e.g. if it's a x16, x8, ... etc. slot.

 

Here are two benchmarks by TPU:

https://www.techpowerup.com/review/nvidia-geforce-gtx-1080-pci-express-scaling/

https://www.techpowerup.com/review/pci-express-4-0-performance-scaling-radeon-rx-5700-xt/

 

Essentially both PCIe generation and lanes have very little impact.