XFX RX-580 GTS Black Edition (RX-580P8DBD6)

[DaBallger]

XFX RX-580 GTS Black Edition (RX-580P8DBD6) (1425OC+/1405TrueCore)

Was able to bench this Gem @ 1500 core/2600 Mem with 1.25Vcore 71c Max Temps.
Core was on the very edge of stability so I backed it down to 1470 for 24/7 use, but the WTF here is the memory clocks.. you ARE seeing it right 2600Mhz (10,400Mhz) up from stock 2000Mhz (8,000Mhz) that's Maxed out in WattMan & Afterburner no Memory errors were checked with HWinfo software during all tests & Benchmarks. This thing is a GEM of a card for the Price/Performance ratio!!! Best card HANDS down for 1080p 144Hz monitors & DX12 gaming @ $260, and only give or take 10-15fps slower then a $400+ GTX 1070 in most situations.. other them synthetic benchmarks which we all know you have to take with a grain of sand for the most part compared to Real World applications and games. This is my 1st RX 580 tested so far, I've tested at least 6 different variations of the RX 480 (Asus STRIX OC was my Fav BTW) this thing FLOORS every RX 480 with ease. Okay, I'm finished .... just my lil in put for now

 

[eidairaman1]

Till it burns up...

 

[DaBallger]

@ 64c I doubt that Check the screen shots

 

[r9]

XFX RX-580 GTS Black Edition (RX-580P8DBD6) (1425OC+/1405TrueCore)

Was able to bench this Gem @ 1500 core/2600 Mem with 1.25Vcore 71c Max Temps.
Core was on the very edge of stability so I backed it down to 1470 for 24/7 use, but the WTF here is the memory clocks.. you ARE seeing it right 2600Mhz (10,400Mhz) up from stock 2000Mhz (8,000Mhz) that's Maxed out in WattMan & Afterburner no Memory errors were checked with HWinfo software during all tests & Benchmarks. This thing is a GEM of a card for the Price/Performance ratio!!! Best card HANDS down for 1080p 144Hz monitors & DX12 gaming @ $260, and only give or take 10-15fps slower then a $400+ GTX 1070 in most situations.. other them synthetic benchmarks which we all know you have to take with a grain of sand for the most part compared to Real World applications and games. This is my 1st RX 580 tested so far, I've tested at least 6 different variations of the RX 480 (Asus STRIX OC was my Fav BTW) this thing FLOORS every RX 480 with ease. Okay, I'm finished .... just my lil in put for now

According this review http://www.overclockersclub.com/reviews/xfx_radeon_rx580_8gb_black_edition/3.htm they could only able to push the mem to 2116MHz.
Not all chips are the same but 500Mhz difference highly unlikely.

 

[uuuaaaaaa]

Would love to see how much it needs to run @ 1266MHz, to compare with the stock rx480's. I smell something like 850mV!

 

[eidairaman1]

According this review http://www.overclockersclub.com/reviews/xfx_radeon_rx580_8gb_black_edition/3.htm they could only able to push the mem to 2116MHz.
Not all chips are the same but 500Mhz difference highly unlikely.

Yup, YMMV

 

[DaBallger]

According this review http://www.overclockersclub.com/reviews/xfx_radeon_rx580_8gb_black_edition/3.htm they could only able to push the mem to 2116MHz.
Not all chips are the same but 500Mhz difference highly unlikely.

 

[cdawall]

My 480 runs a 1480 core just fine. Even with the pair in xfire I was pushing 1450, but am out of psu.

 

[Kanan]

Do benchmarks with 2000 MHz, 2200, 2400 and 2600, so we can be sure if it's working or not.

 

[KainXS]

Try oclmembench and see if the bandwidth drops as you overclock, if it does then you won't see errors.

 

[buildzoid]

I recently saw some one have a glitch on an R9 290 where they could set 2000MHz mem clock. The performace was exactly the same at as 1250MHz mem clock but all the software said mem clock was 2000MHz. 1500MHz was faster than the "2000"MHz. So I would really recomend you test 2000 2100 2200 2300 2400 2500 and 2600 to see if the mem clock doesn't just glitch out past a certain point.

 

[Kanan]

It's probably defaulting to 2000 MHz anyway, I don't believe he has secretly installed G5X on his RX 580.

edit: stupid mistake is stupid, but anyway, I don't think it's ultra high clocking GDDR5 he has

 

[cdawall]

I recently saw some one have a glitch on an R9 290 where they could set 2000MHz mem clock. The performace was exactly the same at as 1250MHz mem clock but all the software said mem clock was 2000MHz. 1500MHz was faster than the "2000"MHz. So I would really recomend you test 2000 2100 2200 2300 2400 2500 and 2600 to see if the mem clock doesn't just glitch out past a certain point.

I wouldn't doubt you are correct on this. Isn't there a physical limitation of the GPU to not exceed 2250 anyway?

 

[buildzoid]

I wouldn't doubt you are correct on this. Isn't there a physical limitation of the GPU to not exceed 2250 anyway?

Nope my RX 480 GTR did 2320MHz on the memory once I moved the 2250MHz BIOS cap to 2500MHz.

 

[DaBallger]

I'm actually quit surprised how many people are immediately denouncing the fact that somebody might of just simply hit the silicon lottery .. that simple. I do understand the "How could that be" mentality, I'm not to sure what or how this thing is hitting those numbers either. I just got back home and will run every bench I have at various clock speeds so maybe I can find out. (Will provide MORE screen shots... but hell it sounds like some people want video or DNA for proof)

 

[Kanan]

Chill its not like we're totally in doubt mode, it's just we want proof because the clocks are a "bit" on the high side.

@buildzoid nice but a far shot from 2600.

 

[cdawall]

Nope my RX 480 GTR did 2320MHz on the memory once I moved the 2250MHz BIOS cap to 2500MHz.

I'll have to play with bios on my cards they pretty easily did 1480/2250. I put a new psu on order to remove my artificial cap as these cards are hungry (255-260w a card according to gpuz, power at the wall matches).

 

[DaBallger]

Quick Update: So after several hours of Bench & test's here are a few highlights. I did notice the scores/performance overall and in DX12 would slowly drop as a raised the Memory clocks up 100MHz at a time starting after 2100MHz thru 2400MHz, but then jumped way back up at 2500-2600MHz. In DX11 test the scores/performance was a steady incline with a slight plateau at 2300-2400 then noticeable increase at 2500-2600Mhz. My overall best scores with thermals, volts, stability, etc. in mind came in around (1475core / 2250mem). I will compile a boatload screenshot's & possibly a video compilation with final results and such. Happy Hunting

 

[DaBallger]

Here is a quick video of Superposition benchmark while running (1475 Core / 2600 Mem) I used AMD's Re-Live to record it and lemme tell you .. not a fan. It DID NOT like any diagnostic screens open and reporting while recording & benching. Anyway enjoy.

 

[Kanan]

We need more numbers than that. Complete benchmark scores for every step (2000 to 2600). Yeah I'm still doubtful

 

[DaBallger]

Just tested & ran a 1500MHz/2600MHz run .. had to pull off a few tricks to make it stable but it passed on air stock cooler. Man I'd love to get a hold of a custom full cover water block for this mudda scooter

 

[DaBallger]

1500/2600 run

 

[eidairaman1]

Just tested & ran a 1500MHz/2600MHz run .. had to pull off a few tricks to make it stable but it passed on air stock cooler. Man I'd love to get a hold of a custom full cover water block for this mudda scooter

Give Alphacool a few months, they might, they seem to do WBs for Custom Cards more so than EK does

 

[RejZoR]

Best way to asses overclock quality is to run 3DMark Stability test and observe the frame quality result in the end. I've had issues achieving 97% target when using wrong OC BIOS parameters for my GTX 980. Or when OC was way too high and I scored like 94-95% range with a failure of course because I didn't met 97% target. Games will run fast, but they might feel choppy or stuttering with such results.

 

[DaBallger]

RejZoR I ran the 3DMark Stability test it scored 99.4% .. Temps did shoot up around 74-75c, but this is the only test that pulls those temps. all other bench & games only hit 64-68c ish