• 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 64 Owners Thread

Soldato
Joined
24 Jul 2004
Posts
22,594
Location
Devon, UK
vega64.jpg
.
 
Man of Honour
OP
Joined
21 May 2012
Posts
31,940
Location
Dalek flagship
How do you know they had faulty memory chips? Did you get a report from the repair/RMA centre, or are you just assuming?

Did the cards only become faulty after you removed the stock thermal solution and fitted waterblocks?

They were faulty on delivery and yes I do know what memory artefacts/faults look like.
 
Soldato
Joined
30 Dec 2011
Posts
5,419
Location
Belfast
are you guys setting state 6 as the minimum state in Wattman when gaming or benching?

No, mine always settles on state 4 or 5 after prolonged use, though it's because I wont let the fans go over 2300 RPM. At this max RPM setting it even goes to state 3 (1138MHz) in Witcher 3 after a short while, where it gives similar performance to my old Fury X. Though for noise and cooling it is a big improvement over the old 290X blowers.

I'm waiting for voltage tweaks to actually work so I can test undervolting. Voltage changes using either WattTool, or Wattman have zero effect with the current bugged drivers.

Hoping for fixed drivers ASAP.
 
Associate
Joined
4 Jun 2016
Posts
243
are you guys setting state 6 as the minimum state in Wattman when gaming or benching?

for benching you need to set the minimum fan speed at like 3k else it will thermal throttle before the fan speeds up; i'm just using balanced atm since the core/voltage tweaks are bugged, when core/volt is fixed i'll probably run the core undervolted with HBM at 1100 until I get a water block!
 
Soldato
Joined
26 Sep 2013
Posts
10,711
Location
West End, Southampton
I've noticed when I don't set state 6 as minimum state, when gaming or benching my core clock graph looks look a sharks set of teeth. It does that regardless of what temperature I'm at. I've seen it dropping to 852mhz even when the temp at that point was 72c.
 
Soldato
Joined
30 Mar 2016
Posts
2,884
Location
East Lothian
Interesting that - I just wonder is it happening all the time or not? Or if this affects Vega 64 in the same manner? Mine certainly behaved as if it was applying the overclocks: I saw noticeable and proportionate gains in scores up to 1907 and 1075 and experienced crashes and artefacts when I went any higher than that. I guess we'll find out when they address it with the next driver release. I'm going to run some benchmarks at stock to compare...

Ok so here are my Time Spy scores, starting from the default Balanced and Turbo profiles and then my overclocks. Each one clearly shows a proportionate performance increase:

Core:1668/1750 (balanced) - HBM: 945 - Total Score: 7130 - Graphics Score: 7490
Core:1750 (turbo) - HBM: 945 - Total Score: 7231 - Graphics Score: 7610
Core:1837 (custom) - HBM: 1050 - Total Score: 7339 - Graphics Score: 7844
Core:1882 (custom) - HBM: 1075 - Total Score: 7479 - Graphics Score: 7872
Core:1907 (custom) - HBM: 1075 - Total Score: 7599 - Graphics Score: 8062

Looking at the graphics score it's a near 8% increase at 1907/1075 compared to stock vs a 9% increase on the core clock. Anything higher on either the core or memory clocks caused Time Spy to either artefact or crash for me. Based on these scores, I don't think I have encountered the bug and believe my overclock is genuine.
 
Last edited:
Associate
Joined
7 Aug 2017
Posts
415
Location
location location
Looking at the graphics score it's a near 8% increase at 1907/1075 compared to stock vs a 9% increase on the core clock. Anything higher on either the core or memory clocks caused Time Spy to either artefact or crash for me. Based on these scores, I don't think I have encountered the bug and believe my overclock is genuine.

Good one - it does seem like you're getting a real overclock. Do you get much variance in score between different runs of the same OC settings?
 
Associate
Joined
16 Aug 2017
Posts
162
Location
Gods own county
Interesting that - I just wonder is it happening all the time or not? Or if this affects Vega 64 in the same manner? Mine certainly behaved as if it was applying the overclocks: I saw noticeable and proportionate gains in scores up to 1907 and 1075 and experienced crashes and artefacts when I went any higher than that. I guess we'll find out when they address it with the next driver release. I'm going to run some benchmarks at stock to compare...

Ok so here are my Time Spy scores, starting from the default Balanced and Turbo profiles and then my overclocks. Each one clearly shows a proportionate performance increase:

Core:1668/1750 (balanced) - HBM: 945 - Total Score: 7130 - Graphics Score: 7490
Core:1750 (turbo) - HBM: 945 - Total Score: 7231 - Graphics Score: 7610
Core:1837 (custom) - HBM: 1050 - Total Score: 7339 - Graphics Score: 7844
Core:1882 (custom) - HBM: 1075 - Total Score: 7479 - Graphics Score: 7872
Core:1907 (custom) - HBM: 1075 - Total Score: 7599 - Graphics Score: 8062

Looking at the graphics score it's a near 8% increase at 1907/1075 compared to stock vs a 9% increase on the core clock. Anything higher on either the core or memory clocks caused Time Spy to either artefact or crash for me. Based on these scores, I don't think I have encountered the bug and believe my overclock is genuine.


According to actual hardcore overclocking who gn get there stuff from when it comes to this you are actually overclocked to that, he says if you are on those clocks you should be hitting around 8k scores, you are so its all good.Its also not just vega that suffers from this, pascal will do it too, software with show increased speeds, your actually worse then stocks which he also has a video on called the pascal problem.
 
Associate
Joined
26 Jun 2015
Posts
410
Interesting that - I just wonder is it happening all the time or not? Or if this affects Vega 64 in the same manner? Mine certainly behaved as if it was applying the overclocks: I saw noticeable and proportionate gains in scores up to 1907 and 1075 and experienced crashes and artefacts when I went any higher than that. I guess we'll find out when they address it with the next driver release. I'm going to run some benchmarks at stock to compare...

Ok so here are my Time Spy scores, starting from the default Balanced and Turbo profiles and then my overclocks. Each one clearly shows a proportionate performance increase:

Core:1668/1750 (balanced) - HBM: 945 - Total Score: 7130 - Graphics Score: 7490
Core:1750 (turbo) - HBM: 945 - Total Score: 7231 - Graphics Score: 7610
Core:1837 (custom) - HBM: 1050 - Total Score: 7339 - Graphics Score: 7844
Core:1882 (custom) - HBM: 1075 - Total Score: 7479 - Graphics Score: 7872
Core:1907 (custom) - HBM: 1075 - Total Score: 7599 - Graphics Score: 8062

Looking at the graphics score it's a near 8% increase at 1907/1075 compared to stock vs a 9% increase on the core clock. Anything higher on either the core or memory clocks caused Time Spy to either artefact or crash for me. Based on these scores, I don't think I have encountered the bug and believe my overclock is genuine.

Yeah but according to haircut 2000 (<--- Credit to Gerard for the nickname lol) if you are seeing higher scores it is because the benchmark is corrupting and not rendering everything... I can't give any personal feedback of course, but I wouldn't be surprised if this not entirely true and I do believe that the 64 LCs are doing 1900+ mhz and rendering correctly, the LC users will have to give some feedback on that. It's just that you have to watch out that the clocks are really being applied, his snarky delivery and clickbaity thumbnail/title is somewhat misleading I think, but time will tell.
 
Soldato
Joined
30 Mar 2016
Posts
2,884
Location
East Lothian
Yeah but according to haircut 2000 (<--- Credit to Gerard for the nickname lol) if you are seeing higher scores it is because the benchmark is corrupting and not rendering everything... I can't give any personal feedback of course, but I wouldn't be surprised if this not entirely true and I do believe that the 64 LCs are doing 1900+ mhz and rendering correctly, the LC users will have to give some feedback on that. It's just that you have to watch out that the clocks are really being applied, his snarky delivery and clickbaity thumbnail/title is somewhat misleading I think, but time will tell.
I did see one incident of that happening; with 10% frequency on the core clock (1917 I think it was), I managed to run Time Spy but you could clearly see there were artefacts and things missing although the overall score dropped to circa 5500. I've watched Time Spy each time I ran it and the scores are definitely legitimate although whilst I've been able to replicate the same scores for the balanced and turbo profiles I'm falling just short of 8000 on the graphics score for my top clock in retesting this morning...

Edit: Ok so I reset the clocks and reapplied it and now I'm hitting 8000+ again on the graphics score after getting a couple of scores falling around 200 short of that. It may be that the clocks aren't applied correctly every time and it's worth resetting and reapplying for multiple tests. Bottom line is that I'm confident Vega 64 can clock to at least 1.9GHz under sufficient cooling.
 
Last edited:
Associate
Joined
26 Jun 2015
Posts
410
Yep it looks like this is what is happening the OCing is buggy and doesn't apply every time, but that isn't a reason to make a "snooty" video saying "NO! your not getting that OC" like never getting those clocks and even if you are it's just not benching properly, it could also be that their modded bios exacerbates this OCing bug in some way. But the delivery of this info (that was pretty much common knowledge before this video) could have been done way better. (Im really beginning to dislike the dude lol not a reason to unsub though of course)

That being said the title isn't incorrect as I haven't seen anyone clock Vega to 2ghz lol
 
Back
Top Bottom