Jump to content

Ryzen 1600 Steam VR high CPU bound frames

My specs are: Ryzen 1600 @3.7ghz, 1.375v

DDR4 16GB Corsair 2666Mhz 16-18-18-35

GTX1060 6GB.

I have never had any problems gaming but when I do the Steam VR performance test I have a super high number of cpu bound frames and I don't know what's causing it. No one else with the same processor seems to have this even if there's is running at a lower clock speed then mine. I have an Oculus rift on the way and am wondering if this will cause problems and if so can I fix it.

Does anyone know what could be causing this? I'm on a fresh install of windows and it's still happening.

5af5923a88da2_Screenshot2018-05-0521_58_41.png.32ff8b80b619c9554c2019223ebd1393.png

Link to comment
Share on other sites

Link to post
Share on other sites

DDR4-2666 CL16... not the best match for a Ryzen CPU. I've said it plenty of times before, Ryzen needs high speed, low latency DDR4 to perform at its best for gaming. You are crippling it significantly by running slower memory, especially slower memory and CL16 timings.

 

Ideally you would want DDR4-3200 CL14, you'll probably get a lot less CPU bound frames that way.

Link to comment
Share on other sites

Link to post
Share on other sites

Well I can't afford to upgrade the ram for a while, I don't know much about overclocking memory so I haven't tried much but the 2993 memory try it! setting works on my motherboard. but gives me only cuts that number down from 12000 to 9800 frames.

Edited by SammWolf
said 1200 instead of 12000
Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, epsilon84 said:

DDR4-2666 CL16... not the best match for a Ryzen CPU. I've said it plenty of times before, Ryzen needs high speed, low latency DDR4 to perform at its best for gaming. You are crippling it significantly by running slower memory, especially slower memory and CL16 timings.

 

Ideally you would want DDR4-3200 CL14, you'll probably get a lot less CPU bound frames that way.

The RAM is OK, that is not really an issue in this case.

 

The SteamVR benchmark is a turd, or at least it used to be last time I checked.

Link to comment
Share on other sites

Link to post
Share on other sites

Try to increase the RAM voltage, most likely it defaults to 1.20V, 1.35V or even 1.4V is fine, you might even get DDR4-3200 speeds, try it and see.

 

FWIW, my Kingston HyperX 2666 can run at up to 3400 CL15 when I give it 1.35V, but that is on an Intel Z370 platform. YMMV on a Ryzen platform.

Link to comment
Share on other sites

Link to post
Share on other sites

1 minute ago, WereCat said:

The RAM is OK, that is not really an issue in this case.

 

The SteamVR benchmark is a turd, or at least it used to be last time I checked.

Could be, all I'm saying is that in actual gameplay testing, Ryzen shows immense performance scaling from faster memory, as well as tight timings.

Link to comment
Share on other sites

Link to post
Share on other sites

Hmm, I'll give it a shot, but I don't know anything about messing with timings.

Link to comment
Share on other sites

Link to post
Share on other sites

Just now, SammWolf said:

Hmm, I'll give it a shot, but I don't know anything about messing with timings.

The key is Cas latency, try CL15-16-16-35, it might even work at stock volts with those settings.

 

More aggressive timings like CL14-14-14-28 for example might need 1.35V+, and even then it might more work, it all depends on the quality of your RAM.

Link to comment
Share on other sites

Link to post
Share on other sites

 

Example of how much performance increases with tuned memory timings (plus higher memory speeds), skip to 7:40 for the gaming benchmarks

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, epsilon84 said:

The key is Cas latency, try CL15-16-16-35, it might even work at stock volts with those settings.

 

More aggressive timings like CL14-14-14-28 for example might need 1.35V+, and even then it might more work, it all depends on the quality of your RAM.

I'll give those timings a go then and see what happens, thanks.

Link to comment
Share on other sites

Link to post
Share on other sites

So I am now running those timings "15-16-16-35 except there's and extra 16, my motherboard is saying there's an extra 16 in there. Speed is still 2666 at the moment as I didn't want to try and change both at once.

I've got memtest but it now says "Memory locking failed (may be reserved by other apps/kernals", it didn't say this when I tested with the old timings.

This change does appear to of made some difference.

5af5a2ca0da33_Screenshot2018-05-1200_01_28.png.8483f7884207006b46a84d39d2d1a1ed.png

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, SammWolf said:

So I am now running those timings "15-16-16-35 except there's and extra 16, my motherboard is saying there's an extra 16 in there. Speed is still 2666 at the moment as I didn't want to try and change both at once.

I've got memtest but it now says "Memory locking failed (may be reserved by other apps/kernals", it didn't say this when I tested with the old timings.

This change does appear to of made some difference.

5af5a2ca0da33_Screenshot2018-05-1200_01_28.png.8483f7884207006b46a84d39d2d1a1ed.png

That's a good start, a 4% improvement from a simple CAS latency tweak from 16 to 15 :)

 

I'm sure higher memory speeds will help also, try increasing the voltage to 1.3V or 1.35V and aim for 2933 at a minimum. That may alleviate the memtest error message too, your RAM may simply require a bit more voltage to run the tighter timings

Link to comment
Share on other sites

Link to post
Share on other sites

I just tried 2933mhz CL15-16-16-16-35 @1.36v and the results actually got worse but nothing crashed or anything, memtest is still saying memory locking failed too.

I forgot to screenshot it but the vr test went from 8.6% cpu bound frames to 13.6%

Link to comment
Share on other sites

Link to post
Share on other sites

Hmmm, that is bizarre, I'm not too family with memtest to be honest, but if its stable I wouldn't worry too much about it.

 

Is there much variance with the VR test between run to run? I'm just wondering what the margin of error is. But performance regression from 2666 to 2933 doesn't make sense to me at all.

Link to comment
Share on other sites

Link to post
Share on other sites

It's Ryzen, ran into this with my 1700X. Had huge issues with AZS. 

If anyone asks you never saw me.

Link to comment
Share on other sites

Link to post
Share on other sites

The variation from run to run isnt huge, I just ran another test now and these are the results.

5af5a8ab2266c_Screenshot2018-05-1200_27_38.png.f222029d35549bdff8db041b227cc962.png

Link to comment
Share on other sites

Link to post
Share on other sites

That benchmark, kinda sucks. Not very representitive. Changes run to run are to be expected. A more accurate benchmark is 3DMarks. 

If anyone asks you never saw me.

Link to comment
Share on other sites

Link to post
Share on other sites

Just now, App4that said:

That benchmark, kinda sucks. Not very representitive. Changes run to run are to be expected. A more accurate benchmark is 3DMarks. 

Yeah but changes that big have never happened in my time running this test, and dropping the timings is the first time it has dropped to that low %, then right after changing memory speed with the same timings it's jumped to high then it was before changing the timings. I've never had any problems with any normal games so my only concern is how VR performance will be affected by this.

Link to comment
Share on other sites

Link to post
Share on other sites

I've noticed chrome tabs that load in the background have started crashing, so this timing/voltage/speed combo isnt working I'm guessing.

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, SammWolf said:

Yeah but changes that big have never happened in my time running this test, and dropping the timings is the first time it has dropped to that low %, then right after changing memory speed with the same timings it's jumped to high then it was before changing the timings. I've never had any problems with any normal games so my only concern is how VR performance will be affected by this.

VR games tend to sit on one or two cores, why Ryzen can have issues. Like I said the stand out for me is Arizona Sunshine which was unplayable with the settings crancked. Runs perfictly with my 8700k using the same graphics card. 

If anyone asks you never saw me.

Link to comment
Share on other sites

Link to post
Share on other sites

Hmm, well ill keep messing with my ram and see if i can get some better settings anyway. But will just have to see how it goes once my rift actually gets here i guess. I have fallout 4 VR, dead effect 2, sairento and VR Chat to test, plus the games bundled with the rift.

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×