R9 Fury Not reaching Expected Speeds

[Dimondminer11]

Ok 1st, I can't change HBM speeds and 2, If I have the clock speed set to 1100 it is only in the ~800mhz range. Is this normal behaviour or no?

[Dimondminer11]

After doing a userbenchmark run it appears as though its reaching max speeds. So this is a Furmark issue.

[Dimondminer11]

Nope I take back the take back. The 2200G can't possibly be holding this card back this hard could it?

[Guest]

Furmark is nasty beating on a gpu. It was probably throttling. 

 

The 2200G is holding the card back yes. I agree to that.

 

let me edit this.

 

It's holding the card back due to clocks.

 

The Userbenchmark is giving you a low score due to low thread count.

 

Maybe a 2600x would boost performance bit. Or a 3600x if you had the coin. 

[Dimondminer11]

6 hours ago, ShrimpBrime said:

Furmark is nasty beating on a gpu. It was probably throttling. 

 

The 2200G is holding the card back yes. I agree to that.

 

let me edit this.

 

It's holding the card back due to clocks.

 

The Userbenchmark is giving you a low score due to low thread count.

 

Maybe a 2600x would boost performance bit. Or a 3600x if you had the coin. 

Ok that's what I was figuring. Thanks!

[Dimondminer11]

Ram speed isn't helping me either. 2400mhz with 14 14 14 14 32 timings. I'll take out the garbage Samsung dimms I just got and oc my skhynix dimms up to 3200mhz and see if I get better performance.

[Guest]

Why is the samsung garbage? 

[Xkillerpn]

1 - Never use furmark on any AMD card , you can kill it 

2 - Dont use userbenq its crap

 

Now to real stuff

 

1 - Make sure you have 2020 drivers, they are very good

2 - Use 3dmark or superposition to test your card, put that power slider to the max and check clocks and temps.

3 -  the 2600x will not hold back the Fury dont worry about it, at this point is a Rx590 performance 

4 - Put windows on high performance as the drivers also on the Tunning part, change to manual and put the power slider at max 

 

hope this helps

[Mark Kaine]

9 hours ago, Dimondminer11 said:

Nope I take back the take back. The 2200G can't possibly be holding this card back this hard could it?

Why yes. Of course it is >.>

 

9 hours ago, ShrimpBrime said:

Maybe a 2600x would boost performance bit. Or a 3600x if you had the coin.

 

Yes they need da coin.

[Mark Kaine]

9 hours ago, Dimondminer11 said:

After doing a userbenchmark run it appears as though its reaching max speeds. So this is a Furmark issue.

Fumark is shit,  who told you to use this crap honestly? 

 

PS : same for userbenchmark... Both applications have very niche use cases,  but at least userbench is quick and doesn't ruin your hardware nilly willy. ~

 

[Xkillerpn]

Just now, Mark Kaine said:

Fumark is shit,  who told you to use this crap honestly? 

 

 

No no no you are worng Furmark is very good to blow up vrms and on AMD card even better were they dont have limit to the voltage specially the older ones  

 

Please no one use this 

[Dimondminer11]

2 hours ago, Xkillerpn said:

No no no you are worng Furmark is very good to blow up vrms and on AMD card even better were they dont have limit to the voltage specially the older ones  

 

Please no one use this 

Jesus I had no idea about Furmark nuking hardware!

[xKr1Sx]

3 hours ago, Xkillerpn said:

1 - Never use furmark on any AMD card , you can kill it 

Why? I think that as long as temps stay under control there shouldn't be any issues. I tested my R9 280 multiple times and nothing happened (maybe I'm lucky lol)

[Zando_]

3 hours ago, Xkillerpn said:

Make sure you have 2020 drivers, they are very good

Yeah naaaaaaaaaaahhhhhhhhhhhh extremely hard pass. They can be very good, but they've also been incredibly shit for a lot of people. 

2 hours ago, Mark Kaine said:

same for userbenchmark... Both applications have very niche use cases,  but at least userbench is quick and doesn't ruin your hardware nilly willy. ~

^^^ Yes, Userbenchmark bad. Use 3DMark, IIRC the standard TimeSpy and FireStrike are free. They're proper benchmarks, and you can compare online to people with that same hardware config (keep in mind the top spots are always people with nutty OCs, not stock parts though). 

 

1 minute ago, xKr1Sx said:

Why? I think that as long as temps stay under control there shouldn't be any issues. I tested my R9 280 multiple times and nothing happened (maybe I'm lucky lol)

Same with my Vega FE, never had issues. Although there's not really a practical reason to use Furmark, it's a power virus. Such an unrealistic load that modern GPUs are designed to notice that and throttle themselves, and it reflects unrealistic temps. Long renders/Valley or so on a loop, or 3DMark runs once the system is warm are a more accurate reflection of real world temps. 

[xKr1Sx]

7 minutes ago, Zando Bob said:

^^^ Yes, Userbenchmark bad. Use 3DMark, IIRC the standard TimeSpy and FireStrike are free. They're proper benchmarks, and you can compare online to people with that same hardware config (keep in mind the top spots are always people with nutty OCs, not stock parts though). 

On steam technically TimeSpy is free, but 3DMark is required and 3DMark's normal price is 30$, sale price is 5$. Or maybe I can get it somewhere else for free? (legit ways)

[Zando_]

2 minutes ago, xKr1Sx said:

On steam technically TimeSpy is free, but 3DMark is required and 3DMark's normal price is 30$, sale price is 5$. Or maybe I can get it somewhere else for free? (legit ways)

https://benchmarks.ul.com/3dmark#windows

 

[xKr1Sx]

5 minutes ago, Zando Bob said:

https://benchmarks.ul.com/3dmark#windows

Thank you! Didn't know about the Download Demo button

[Zando_]

Just now, xKr1Sx said:

Ah, thank you! Didn't know about the Download Demo button

I didn't either lol, used to be able to just download it from their site. I bought it when it moved to Steam lol, was simpler and I do actually go for bench scores sometimes. 

[Dimondminer11]

17 minutes ago, Zando Bob said:

I didn't either lol, used to be able to just download it from their site. I bought it when it moved to Steam lol, was simpler and I do actually go for bench scores sometimes. 

I did try running it. It ran on the system without a dGPU but it now doesn't seem to be able to get system info.

[Dimondminer11]

On 1/7/2020 at 12:22 PM, Zando Bob said:

https://benchmarks.ul.com/3dmark#windows

 

Ok so I ran benchmarks and got a graphics score of 4776 and a cpu score of 3210. Is that about on par to what I should be expecting? On time spy.

[KnowledgeByTech]

On 1/6/2020 at 10:02 PM, Dimondminer11 said:

Ok 1st, I can't change HBM speeds and 2, If I have the clock speed set to 1100 it is only in the ~800mhz range. Is this normal behaviour or no?

perfectly normal, its lower speeds are based on utilization (lower utilization = lower clocks), you can set the minimum and maximum clock speeds more intricately in the new Radeon Wattman, whereas apps like afterburner just raise your maximum allowed clock (at 90%+ utilization).

[Dimondminer11]

9 hours ago, KnowledgeByTech said:

perfectly normal, its lower speeds are based on utilization (lower utilization = lower clocks), you can set the minimum and maximum clock speeds more intricately in the new Radeon Wattman, whereas apps like afterburner just raise your maximum allowed clock (at 90%+ utilization).

Oh cool! Thank you

[Zando_]

10 hours ago, Dimondminer11 said:

Ok so I ran benchmarks and got a graphics score of 4776 and a cpu score of 3210. Is that about on par to what I should be expecting? On time spy.

Seems to be totally normal for that combo. Only 15 runs have actually been done of TimeSpy, but yours is on par with all the rest: