Question what clock speed shoild i keep on my rx 570 to make it as equal as rx 580?

plz tell the exact frequency i should increase
There are no exact settings with overclocking. It’s trial and error, you make a small increases and test it, then if successful repeat until you find the best stable overclock. You could have 2 systems with the same hardware and they overclock differently. I’d recommend Googling gpu overclocking guides and reading/watching a few first. They will be more detailed than what someone can type here.
 
Sep 7, 2020
38
0
30
0
There are no exact settings with overclocking. It’s trial and error, you make a small increases and test it, then if successful repeat until you find the best stable overclock. You could have 2 systems with the same hardware and they overclock differently. I’d recommend Googling gpu overclocking guides and reading/watching a few first. They will be more detailed than what someone can type here.
ok thx
 

tennis2

Honorable
plz tell the exact frequency i should increase
Assuming you can get your VRAM to run at 2000MHz, you'd be looking at a 1430MHz core clock to match stock RX580 performance.

You're unlikely to get that high using "auto" voltage and/or offsets, since you'll hit power limits far before that. You'll need to manually adjust the frequency/voltage curve of the performance states to conserve as much power headroom as possible. You can do this in
  • AMD Settings - Performance - Tuning
  • Tuning Control = Manual
  • GPU tuning = enabled
  • Advanced control = enabled (this will expose the frequency and voltage of each performance state)
  • Voltage = manual (this will allow you to change the voltage for each performance state frequency)
  • VRAM tuning = enabled
  • Advanced control = enabled
  • Power Tuning = enabled (you'll probably want to max this out)
The VRAM voltage acts as a lower limit to the core voltage also, since the core and VRAM get the same voltage. It prevents the situation where the core frequency (and corresponding voltage) can throttle down to a level below the minimum stable voltage for the VRAM. Try 950mV for this and see where that gets you. Essentially then, any core frequency with a corresponding voltage <950mV (generally 1200Mhz and below) will just receive 950mV. Your idle state (State 0) voltage will still be preserved though.

I'm able to hit 1440MHz @ 1175mV (max for my GPU) on my RX480
 
Last edited:
Sep 7, 2020
38
0
30
0
Assuming you can get your VRAM to run at 2000MHz, you'd be looking at a 1430MHz core clock to match stock RX580 performance.

You're unlikely to get that high using "auto" voltage and/or offsets, since you'll hit power limits far before that. You'll need to manually adjust the frequency/voltage curve of the performance states to conserve as much power headroom as possible. You can do this in
  • AMD Settings - Performance - Tuning
  • Tuning Control = Manual
  • GPU tuning = enabled
  • Advanced control = enabled (this will expose the frequency and voltage of each performance state)
  • Voltage = manual (this will allow you to change the voltage for each performance state frequency)
  • VRAM tuning = enabled
  • Advanced control = enabled
  • Power Tuning = enabled (you'll probably want to max this out)
The VRAM voltage acts as a lower limit to the core voltage also, since the core and VRAM get the same voltage. It prevents the situation where the core frequency (and corresponding voltage) can throttle down to a level below the minimum stable voltage for the VRAM. Try 950mV for this and see where that gets you. Essentially then, any core frequency with a corresponding voltage <950mV (generally 1200Mhz and below) will just receive 950mV. Your idle state (State 0) voltage will still be preserved though.

I'm able to hit 1440MHz @ 1175mV (max for my GPU) on my RX480
which software do u benchmark? to check stability? i use kombustor fr 15 mins...is it enough? but when i use heaven benchmark, then it crashes in 5 mins...why?should i stick with kombustor?
 

tennis2

Honorable
Different tests vary on how strenuous they are and (potentially) what aspects are stressed more than others.

I've used FurMark, F@H, and OCCT.
  • OCCT is INCREDIBLY strenuous, probably the most strenuous I've seen. Power consumption is through the roof, so you generally hit power limits very early/low on the frequency testing.
  • F@H is a great "tough-but-not-unrealistic" load that susses out problematic settings relatively quickly. In my experience, you get pretty stable settings out of this, but you might still run into a couple hitches here and there.
  • FurMark is the least strenuous. It's fine for hammering out some rough settings.
 
Sep 7, 2020
38
0
30
0
Different tests vary on how strenuous they are and (potentially) what aspects are stressed more than others.

I've used FurMark, F@H, and OCCT.
  • OCCT is INCREDIBLY strenuous, probably the most strenuous I've seen. Power consumption is through the roof, so you generally hit power limits very early/low on the frequency testing.
  • F@H is a great "tough-but-not-unrealistic" load that susses out problematic settings relatively quickly. In my experience, you get pretty stable settings out of this, but you might still run into a couple hitches here and there.
  • FurMark is the least strenuous. It's fine for hammering out some rough settings.
furmark is my nightmare!😂 that thing will torture my gpu not stress test!
 

mjbn1977

Honorable
Aug 20, 2015
666
47
11,240
53
You will not make it the same speed (otherwise the chip would have ended in a rx580) You can maybe close the gap a little, but if you had a rx580 you also could overclock that one .....so the gap stays the same, kinda. So, you should see this more as an attempt to get a little extra performance out of the rx570, not so much as an attempt to match rx580 performance.
 

tennis2

Honorable
You will not make it the same speed (otherwise the chip would have ended in a rx580) You can maybe close the gap a little, but if you had a rx580 you also could overclock that one .....so the gap stays the same, kinda. So, you should see this more as an attempt to get a little extra performance out of the rx570, not so much as an attempt to match rx580 performance.
Trying to match "stock" RX580 performance.
 
Sep 7, 2020
38
0
30
0
You will not make it the same speed (otherwise the chip would have ended in a rx580) You can maybe close the gap a little, but if you had a rx580 you also could overclock that one .....so the gap stays the same, kinda. So, you should see this more as an attempt to get a little extra performance out of the rx570, not so much as an attempt to match rx580 performance.
ok, but i am trying to match it to stock rx 580 performance, not oced rx 580. just see this video, it defeated stock rx 580
View: https://youtu.be/lgdYeXBr0Mg
 
Sep 7, 2020
38
0
30
0
Trying to match "stock" RX580 performance.
ok, so i have a question...a question which i had in my mind from many days. so i just oced my rx 570. i dont kow why! first i stress tested it on kombuster. temps r 83c which is ok..but then while playing games my temps were wayyy tooo low..i dont know why! i mean, gpu usage is 100% in all aspects, all clock speeds at max...then why there is a temp difference?? plzzz let me know...and at furmark i get 90+ temp😂 lol...at max load there r temp differences...explain this plz .
 

mjbn1977

Honorable
Aug 20, 2015
666
47
11,240
53
because those stress test are specially made in a way to really use the GPU chip to its maximum. Most games will never even come close to stress the GPU that much. 100% utilization in one application/game is not equal to 100% utilization in another application. It's kinda hard to explain but the GPU can process different kind of instruction sets. Simplified, instructions a hardcoded possible calculation a GPU (or CPU) can calculate per clock cycle and thread. Depending what the application needs calculated in a specific situation, certain instructions will be executed. Some instructions are more complex than other and use more transistors than others. A stress test is synthetically created in a way that the GPU will get stressed with the most complex and demanding instructions and thus create more heat. Games usually use a more balanced mix of instructions or some of the instructions not at all.

Again, I explained that very simplified to give you the general idea. An electrical engineer can probably explain that much better.
 
Sep 7, 2020
38
0
30
0
tha
because those stress test are specially made in a way to really use the GPU chip to its maximum. Most games will never even come close to stress the GPU that much. 100% utilization in one application/game is not equal to 100% utilization in another application. It's kinda hard to explain but the GPU can process different kind of instruction sets. Simplified, instructions a hardcoded possible calculation a GPU (or CPU) can calculate per clock cycle and thread. Depending what the application needs calculated in a specific situation, certain instructions will be executed. Some instructions are more complex than other and use more transistors than others. A stress test is synthetically created in a way that the GPU will get stressed with the most complex and demanding instructions and thus create more heat. Games usually use a more balanced mix of instructions or some of the instructions not at all.

Again, I explained that very simplified to give you the general idea. An electrical engineer can probably explain that much better.
thank you so much!
 

madmatt30

Titan
Ambassador
You can't match a 'stock' rx 580 unless you've got a golden card with ridiculous overclocking headroom.

The 580 has 12% more stream processors hardware wise.

Which means you'd have to be hitting 1500mhz+ on the core to get even close.
That would be a 20% overclock, I doubt that's even close to possible.
 

ASK THE COMMUNITY

TRENDING THREADS

Latest posts