sub par performance from R7 1700 and GTX 1070

[F1tch]

synthetic benchmarks give good results, however when in real world application on games (pubg) for example i get maybe 30% cpu usage and my gpu hovers around 50-75% i suffer huge frame drops lag and all the other nastiness that comes with,i am gaming at 1080p but i dont see how its a cpu bottlekneck when i only have a max of 30% usage.

its interesting to note that on my previous 1800x paired with the same graphics card my cpu would share the processing load evenly there was never a single core that maxes out at 100% however on my 1700 this happens frequently i have set my pc to high performance power plan i have edited the threshold for # parked cores making it 0 cores to be parked at any time the only functional difference between my current system and my previos one is that the 1800x system had a shite mobo like the cheapest of cheap i think it cost me about £38 where as my current one is a lot more OC friendly and i now have an m.2 SATA SSD for my OS drive instead of a bog standard SSD the ram speed in both cases is 2133, though currently i have oc'd my ram to 2933MHz and that seemed to allieviate my problem a little however my previous system ran smooth as silk without the bump in ram speeds

current specs:
R7 1700x @ 4GHz@1.413V  (3.93 effective because for whatever reason my bus speed sticks at 99.8)

ASUS PRIME X370-PRO

16GB (2x8GB) of corsair vengance lpx 2133 rated (oc to 2933 with a cl of 17 i think but dont quote me)

ASUS GTX 1070 8GB DUAL OC

WD green M.2 120GB SATA
WD Black 7500RPM 1TB
WD Blue 1TB

 

temps are well within threshold despite using the wraith spire my r7 came with (max of 62c with aggressive fan speed and management under full load) and i see no thermal throttling of any kind

i have disabled every option i can find in the bios for anything that will reduce performance and after all the fiddling and testing i can think of i cannot nail down why im suffering these issues

 

the 1700 and the 1800x are the same chip same silicone same factory line just one has been binned for a higher TDP so i dont think its inherent to the chip i just for the life of me cant figure out wht is causing this issue

 

any help would be greatly appreciated, if you require dxdiag or anything else dont hesitate to ask

 

 

 

[F1tch]

Bump! or is this in the too hard to deal with pile?

[Emberstone]

Make sure you're on the Ryzen Balanced power plan as that's been known to affect load distribution in certain scenarios. Are you trying other games besides PUBG? That's a game that isn't particularly fond of AMD processors for whatever reason.

[F1tch]

what i dont understand is why it wasnt showing this behaviour before and why it is now and most other games are similar just less demanding so its less noticable, also i wouldnt know where to find the ryzen balanced power plan im running a desktop with 0 amd software on it being set to high performance removes core parking in effect and should produce better scalability

[F1tch]

the most baffling thing is while i am in the menu for pubg my gpu gets 100% utilisation the second i get ingame it drops to 40-50% if i change my ingame resolution to 1440p using nvidia dsr i immediately get 99% utilisation

[Emberstone]

5 hours ago, F1tch said:

the most baffling thing is while i am in the menu for pubg my gpu gets 100% utilisation the second i get ingame it drops to 40-50% if i change my ingame resolution to 1440p using nvidia dsr i immediately get 99% utilisation

You should quote people so we know you replied, please. We get notifications when you do.

 

Also, you mentioned previously that you have no AMD software installed. You should get your chipset drivers installed, then.

 

Additionally, the difference in GPU utilization between PUBG's menu and the game itself is that there's no game logic to be processed in the menu, so it can just go wild sending render calls to the GPU (e.g. no CPU limitations), but in-game that changes.

 

As mentioned before, PUBG isn't particularly fond of Ryzen, and especially doesn't like first gen Ryzen. You might be feeling that a little bit. Did you play it on your 1800X, and did it perform differently?

 

Get your chipset drivers installed from your motherboard manufacturer and use the Ryzen Balanced power plan that comes with it. See if that helps out.

[Columbo]

Pubg is a broken game and you shouldn't concern yourself with any benchmarks from that game. 

[Reznik]

If you don't see AMD Ryzen Balanced power plan in the same spot you set it to high performance then you don't even have the Chipset drivers installed.  Start with that.  

[F1tch]

17 hours ago, Emberstone said:

You should quote people so we know you replied, please. We get notifications when you do.

 

Also, you mentioned previously that you have no AMD software installed. You should get your chipset drivers installed, then.

 

Additionally, the difference in GPU utilization between PUBG's menu and the game itself is that there's no game logic to be processed in the menu, so it can just go wild sending render calls to the GPU (e.g. no CPU limitations), but in-game that changes.

 

As mentioned before, PUBG isn't particularly fond of Ryzen, and especially doesn't like first gen Ryzen. You might be feeling that a little bit. Did you play it on your 1800X, and did it perform differently?

 

Get your chipset drivers installed from your motherboard manufacturer and use the Ryzen Balanced power plan that comes with it. See if that helps out.

yes on my 1800x it performed properly the CPU load was as even as i've ever seen it in any game and utilized all of my cores even though i had underclocked my 1800x because at the time i didnt know about the massive temperature offset, i also downloaded the chipset drivers from my motherboard provider and installed them when i first put this machine together.
 

 

14 hours ago, Columbo said:

Pubg is a broken game and you shouldn't concern yourself with any benchmarks from that game. 

im not using pubg to benchmark anything, what concerns me is the odd load sharing between cores that is present in every demanding application that i use that isnt a synthetic benchmark including OBS (for live broadcasting or video recording, i only ever use CPU encoding because i only have one GPU)
 

ill try re-installing the chipset drivers and see if that helps thanks for the speedy replies and sorry for not quoting you in my original replies

 

[F1tch]

3 hours ago, F1tch said:

yes on my 1800x it performed properly the CPU load was as even as i've ever seen it in any game and utilized all of my cores even though i had underclocked my 1800x because at the time i didnt know about the massive temperature offset, i also downloaded the chipset drivers from my motherboard provider and installed them when i first put this machine together.
 

 

im not using pubg to benchmark anything, what concerns me is the odd load sharing between cores that is present in every demanding application that i use including OBS (for live broadcasting or video recording, i only ever use CPU encoding because i only have one GPU)
 

ill try re-installing the chipset drivers and see if that helps thanks for the speedy replies and sorry for not quoting you in my original replies

 

having re-installed the chipset drivers this has made 0 difference to my current issue, personally i think its something to do with the way tasks are assigned to the cores by windows but having used exactly the same ISO of Win 10 as my previous 1800x build there should be no deviation its why im at such a loss, i think by this point its either the SMU is not properly delegating tasks or Windows 10 isnt i for one really hope its the latter not the former and still leaves me at a loss as to how to rectify this issue

 

[Emberstone]

5 hours ago, F1tch said:

having re-installed the chipset drivers this has made 0 difference to my current issue, personally i think its something to do with the way tasks are assigned to the cores by windows but having used exactly the same ISO of Win 10 as my previous 1800x build there should be no deviation its why im at such a loss, i think by this point its either the SMU is not properly delegating tasks or Windows 10 isnt i for one really hope its the latter not the former and still leaves me at a loss as to how to rectify this issue

 

Just to clarify, the Ryzen Balanced power plan is on, correct?

 

Additionally, is this a clean install of Windows, or is it moved from your other computer?

[F1tch]

 

2 hours ago, Emberstone said:

Just to clarify, the Ryzen Balanced power plan is on, correct?

 

Additionally, is this a clean install of Windows, or is it moved from your other computer?

brand new install from pre existing usb bootable, yes the AMD Ryzen balanced power plan is on

[Emberstone]

2 hours ago, F1tch said:

 

brand new install from pre existing usb bootable, yes the AMD Ryzen balanced power plan is on

Man, I'm at a loss then. I've heard of cores being disabled on a fresh install just due to Windows weirdness (can be fixed in Windows easily), but never heard of strange load distribution that wasn't fixed via the Ryzen Balanced plan.

 

I've owned two Ryzen machines now (R5-1600X and Ryzen 7 2700U laptop), and both distributed load as expected for the software being run. PUBG used around six threads, Overwatch used around 12 threads, Doom showed a nearly perfect uniform load, and older games like Starcraft Remastered and Heroes of the Storm showed an appropriate 1 thread load.

 

And all was the same on my 8700K rig.

 

What kind of Cinebench R15 scores do you get?

[F1tch]

38 minutes ago, Emberstone said:

Man, I'm at a loss then. I've heard of cores being disabled on a fresh install just due to Windows weirdness (can be fixed in Windows easily), but never heard of strange load distribution that wasn't fixed via the Ryzen Balanced plan.

 

I've owned two Ryzen machines now (R5-1600X and Ryzen 7 2700U laptop), and both distributed load as expected for the software being run. PUBG used around six threads, Overwatch used around 12 threads, Doom showed a nearly perfect uniform load, and older games like Starcraft Remastered and Heroes of the Storm showed an appropriate 1 thread load.

 

And all was the same on my 8700K rig.

 

What kind of Cinebench R15 scores do you get?

the really fucked up thing is im getting good cinebench and heaven scores my 2 cinebench runs were 1674 for my first then i read that the SOC voltage also effects things like your i/o controller and other bits not just your memory controller so i knew my previous SOCV was 0.9V as auto in the bios i decided to bump it to 1.1V and the cinebench score i just got was 1740 my ram is 2133MHz and i know that impacts performance a little but i had the same speed ram in my 1800x system and it showed none of these issues given they are exactly the same chip you should see (sillicone lottery dependent) the same performance if they are both set to run at the same clock speed

 

EDIT: i've just realised for the second cinebench run i also upped my ram speed to 2666 my ram isnt rated for it though and im a novice when it comes to ram overclocking so i set the speed and let the motherboard oc the rest so my timings are probably toss but it seems to have made a small difference now when i run pubg i get 30%-50% gpu usage which is my current problem but im also seeing cpu usage now jump up to a high of 36%

unigene heaven score was a 2246 and it utilised all of my cores properly and maxed out my gpu with only one frame drop during animation 4

 

games dont utilize the core properly in pubg i get 2 cores high utilisation and the rest are toss
 

doom on the vulkan api runs like hot grease on almost any system especially so if you have a decent amount of threads and a good gpu

 

EDIT 2: single core scores were 157 for the former and 161 for the latter still doesnt beat out an i7-4770k which is dissapointing id expect AMD's IPC to have come a little further than that