• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

The RX Vega 56 Owners Thread

WOW....best I can get with 1800X @ 3.9Ghz ram @ 333MHz and Vega with 1730MHz solid core speed and mem 1050Mhz is 4380 at 1080p Extreme preset (HBCC makes no difference) ......Vega ram speed between 950 - 1050 make negligible difference so I'm just wondering how you can get 4551 with a 1558 core speed :confused:


If you look here you will see no one is getting those scores with Extreme preset regardless of clocks. https://docs.google.com/spreadsheets/d/1qBRTNCluCB6wTBqCyzbWjIPLa2d2BKL-mPCc48-MWKg/edit#gid=0
That's weird, my avg fps is 34, with min of 27, idk why everyone is getting low scores.
 
You sure that was the 1080p Extreme preset and not the High preset???....score seems a little too high for the extreme preset considering your clocks.

Definately 1080p extreme

Ok can confirm that HBCC set to max makes a difference....just got 5000 with stock clocks of 1556 with the 46 AIO Liguid bios

My score was with HBCC set at the lowest amount, with it off I only got 4022 at same oc speeds. Will see if setting it to max improves my score further tonight :)

Was going to ask if anyone had tried the 64 liquid cooled bios. Will stick it on my card tonight. Is there any difference between the liquid and air 64 bios' other than then bump up in mhz?
 
Definately 1080p extreme



My score was with HBCC set at the lowest amount, with it off I only got 4022 at same oc speeds. Will see if setting it to max improves my score further tonight :)

Was going to ask if anyone had tried the 64 liquid cooled bios. Will stick it on my card tonight. Is there any difference between the liquid and air 64 bios' other than then bump up in mhz?

Yes I'm using the AIO Liquid bios and the power tables are different :)
 
Glad to see HBCC helping, is 32gb over 16gb making any difference or an advantage doesnt run to that extent.
One game I tried so far on max settings, oversampling and HBCC on, 24gb isnt enough for it. But that was possibly because the game itself is caching so HBCC in addition was too much
 
Thought I would have a quick go at this Firestrike to see how the system is performing, I'm not really using it for that much gaming but wanted to check that it was about right:



1950x @ stock
vega 56 on its stock bios, with +50% limit, no messing about with voltages but hbm knocked up to 950, HBCC Off.
64gb GSKill RGB 3466 @ 2933 16-16-16-36 1T

Had a look back in this thread and it doesn't look too bad but would probably benefit from some more fiddling tbh.
 
With 16gb system ram what should I allocate HBCC to?

Also I noticed I get 1k higher score on fire strike with hbm set to 900mhz vs 920mhz. I believe it's due to temperature. The low hbm speed means lower overall temp for mem and gpu as mem temp effects gpu temp.
 
With 16gb system ram what should I allocate HBCC to?

Also I noticed I get 1k higher score on fire strike with hbm set to 900mhz vs 920mhz. I believe it's due to temperature. The low hbm speed means lower overall temp for mem and gpu as mem temp effects gpu temp.
amdmatt mentioned using the lowest value for 16gb which is what I did. No issues at this end, scores went up with it on.
 
Just finished strapping a NZXT Kraken X41 to my 56 with a G10 (straight off the R9 290) - required redrilling both plates and it will be very difficult to effectively heat sink the closest MOSFETS using this mod. It has however taken mine from throttling at stock clocks in superposition to hitting a max of 32c on the GPU during the bench and idle sits at 19c according to GPU-Z. I think that's a bit off but with the stock blower it idled at about the 32 mark.

Just got to get my head around wattman now - never used it before. Going to undervolt and see how that goes.
 
idle sits at 19c according to GPU-Z

Unlikely, if the card really idled at 19C your room temperature would have to be at least several degrees below that. It's more likely it reached some threshold it can no longer really monitor, and/or GPU-Z is simply wrong in calculating it.
 
Ambient is 11 tonight. It's probably not right like I said, but it's not below it and it's a fair bit better than it was inaccurately reporting before the mod. it was throttling at 74 previously so the software was keeping pace.
 
Just finished strapping a NZXT Kraken X41 to my 56 with a G10 (straight off the R9 290) - required redrilling both plates and it will be very difficult to effectively heat sink the closest MOSFETS using this mod. It has however taken mine from throttling at stock clocks in superposition to hitting a max of 32c on the GPU during the bench and idle sits at 19c according to GPU-Z. I think that's a bit off but with the stock blower it idled at about the 32 mark.

Just got to get my head around wattman now - never used it before. Going to undervolt and see how that goes.

I did a similar mod on my Vega 64 air and temperatures at stock (220W) are 45c load , idle 23c. Undervolting to around .900mV voltage gives ~42c load and 23c idle. This is with 2x Be Quiet Wings 2 fans at 1000RPM. What temperatures is Wattman reporting on yours, and what fan speed on the radiator fan? Honestly 32c seems very low, great if accurate :)
 
Score 365, GPU 6850 @849/1100, CPU Q9550 @3.81, megatron
Score 424, GPU 6850 @1030/1200, CPU 4690k @4.5, megatron
Score 1001, GPU 280X @1150/1600, CPU 4690k @4.5, megatron
https://forums.overclockers.co.uk/posts/31163348
Same CPU and all, new RX vega 56. My first dabble at increasing the settings after installing OcUK Techlabs AIO GPU cooler, settings garnered from watching youtube re. 1100mv state 6/7 and 950 on memory. Its supposed to be good for 1050-1100 memory with the 64 bios.

https://imgur.com/a/9mRax

Score 2234, GPU RX Vega 56 @1580/950, CPU 4690k @4.5, megatron
DpoIQ

DpoIQ
Finally got it back together. I've got a GPU to 4pin PWM cable into a y-splitter. The sence signal is on the G10 mounted fan. The other side of the y-splitter is connected to a loom of 5 pwm splitters connected to the 4 rad fans and a fifth directed at the other side of the mosfets.

I've been also working on getting my 280x into a saleable condition with stock heatsink + bought 2 packs of 4mm high ramsinks but only used 1. I modded a single heatsink to go between the heatpipes and I'll have to put some akasa thermal gap filler for the rest cause I don't feel like sanding 10 copper 4mm sinks to 2mm.

Edit:
First thing I tried was 3dmark and got a reboot. Then played eternal crusade (f2p got steam workshop now too). Then Unigine benchmark. Using 17.9.1.
 
Last edited:
I did a similar mod on my Vega 64 air and temperatures at stock (220W) are 45c load , idle 23c. Undervolting to around .900mV voltage gives ~42c load and 23c idle. This is with 2x Be Quiet Wings 2 fans at 1000RPM. What temperatures is Wattman reporting on yours, and what fan speed on the radiator fan? Honestly 32c seems very low, great if accurate :)
Just switched on, wattman reports 15c @ idle - GPU-Z agrees on the core, hot spot 17c on GPU and HBM 21c.

I don't have the fans on control at this point - they are just case fans as far as the system is concerned. I have the X41 on push / pull and they aren't noisy but not silent. They've never bothered me enough to warrant lowering the RPM and that might be why I'm getting lower temps.

EDIT. Just ran superposition in 1080 extreme, max GPU 27c, hot spot 42c, HBM 27c. It's a cool morning but not cold, probably 12c or so. This is all stock settings in balanced mode, pretty happy with the temps.
 
Last edited:
Back
Top Bottom