-
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
- Thread starter Kaapstad
- Start date
Tried an old driver, still the same thing. My last chance would be to try the second psu cable. From there on, If I still have problems, I will probably RMA the gpu, don't think it's worth buying a new psu, since the newer gpu's (rtx 20xx/5700) would work without any problem I guess. To be honest I didn't think seasonic would have such problems, I bought it in december 2017 and it was by far (at that time, probably today as well) the best selection according to the internet discussions. I can't believe people from the other forum use the same gpu without any problems with corsair, super flower psu... which all are between 550-600w, so -1 for seasonic from me.
Soldato
I dont know about that. They admitted its affecting the cards so if you have a Seasonic then you need to rule it out by trying another as it may not be the GPU at all. In the detail it states too much noise on the power and it impacts the GPU.
I have a Seasonic, but its an older model 750w and it works fine, even had another GPU in a different slot using both no issues.
I have a Seasonic, but its an older model 750w and it works fine, even had another GPU in a different slot using both no issues.
Even though you might have sent the card back, pull out the PCIe power cables from the PSU and examine them thoroughly. I had similar issues once with my Vega 64 even after switching to 2 independent PCIe 8x cables. Turned out that on my first run with the card using 1 cable with splitter that the casing that goes into the PSU had a pin melted. So when switched to 2 independent cables i reused the one already plugged in (with burn damage). I had to replace that with a new cable and use another new cable before my issues went away. If you have a damaged plug the problem will just continue no matter the card.
Soldato
Interestingly I had the same thing very recently with my Asus ROG Strix RX OC Edition Vega 64. Games would appear to crash but sound was playing fine. Monitor went black sometimes and the power led flashed. Things progressively became worse and it would happen when in Windows. Then I couldn't even launch games (black screen). My PSU was a 700w OCZ jobby, over 10 years old and I've only had the GPU for about a year and a half. So for me it was a toss up between GPU RMA or PSU replacement... and I had planned to change the PSU, so I took a punt with that (thinking the GPU was still fairly new).
I bought the Seasonic 1300w Prime Gold PSU. Problem solved! As people have said, you should run two seperate cables, don't use one with the splitter. PC is rock solid now so it was the old PSU giving up the ghost.
Hope this helps, but seeing your symptoms I thought I best reply.
Today because I had some time I tried different things. As you may know, I was powering the gpu with 2 separated cables and still had problems.
1. I disconnected cable 1 and left only cable 2 (both 8pin connectors in the gpu)
2. My modular seasonic has 4 inputs x 8 pins for CPU/PIC-E cables. I disconnected the first 8 pin cable from the psu (near the mobo one) and connected there the gpu cable 2. I moved the disconnected cable to the last position.
3. I forced in bios gen3 for pcie.
Currently I have 24 hours of pc usage without any black screens, usually I had them in 1 hours or less... I still don't get way. And yes, currently there is only one cable connected to the gpu. Is there something like a faulty cable
1. I disconnected cable 1 and left only cable 2 (both 8pin connectors in the gpu)
2. My modular seasonic has 4 inputs x 8 pins for CPU/PIC-E cables. I disconnected the first 8 pin cable from the psu (near the mobo one) and connected there the gpu cable 2. I moved the disconnected cable to the last position.
3. I forced in bios gen3 for pcie.
Currently I have 24 hours of pc usage without any black screens, usually I had them in 1 hours or less... I still don't get way. And yes, currently there is only one cable connected to the gpu. Is there something like a faulty cable
Soldato
Trace it back to the PSU port. I have had a poor socket on one of them modular ones before, that was RMA'd. Yeah the cables can get damaged or burn back if theres not a great contact.
An update: usually I consider RMA's a hassle and so, I did bet and bought a new psu, 2 weeks have past and no problems, so far. Currently I run as OC/UV:
P6: 1550 Mhz /1000Mv
P7: 1680/1100
Memory: 1100/980
I get 25.5k in Firestrike. A friend of mine bought a medium 2070 super and he has the same results, so far, I'm happy with it. The cost of the psu was really not expected by me.
https://www.corsair.com/eu/en/Categ...™-80-PLUS-Gold-Power-Supplies/p/CP-9020195-EU
P6: 1550 Mhz /1000Mv
P7: 1680/1100
Memory: 1100/980
I get 25.5k in Firestrike. A friend of mine bought a medium 2070 super and he has the same results, so far, I'm happy with it. The cost of the psu was really not expected by me.
https://www.corsair.com/eu/en/Categ...™-80-PLUS-Gold-Power-Supplies/p/CP-9020195-EU
Soldato
Glad its now working as intended for you!
Some returned the GPU in this situation. Must be thousands of heartbroken Vegas in warehouses due to poor diagnosis.
Yes, what will be next:
1. buy another set of 16gb ram in order to activate HBCC (even if the gain is so small), it might help I guess in games like Call of Duty, where the VRam usage sometimes is crazy and leaking from time to time. It was a shock also, that the same ram kit was 115 gbp in 2017, now it's 68 and I see a good opportunity to upgrade from 16 to 32 and call it an end for this build (till ddr5 will arrive)
2. I still have bottleneck in many games. I overclocked my ryzen 1600 to 3.8 but I didn't see too much improvement as gpu usage. There so, I will probably buy a ryzen 3600 and a CM MasterLiquid AIO to update my current CM 212 cooler.
1. buy another set of 16gb ram in order to activate HBCC (even if the gain is so small), it might help I guess in games like Call of Duty, where the VRam usage sometimes is crazy and leaking from time to time. It was a shock also, that the same ram kit was 115 gbp in 2017, now it's 68 and I see a good opportunity to upgrade from 16 to 32 and call it an end for this build (till ddr5 will arrive)
2. I still have bottleneck in many games. I overclocked my ryzen 1600 to 3.8 but I didn't see too much improvement as gpu usage. There so, I will probably buy a ryzen 3600 and a CM MasterLiquid AIO to update my current CM 212 cooler.
Don't bother with the MasterLiquid. Your CM 212 is plenty. I've just upgraded myself to 32gigs everything is butter smooth with HBCC on. My 3600 is cooled by a cryorig R1 ultimate and that cooler is total overkill even at 4,3ghz and 1.43volts. Stresstesting only sees the CPU at around 69-70 degrees so your CM 212 cooler will be fine for regular use and this is coming from someone who had a 8350 overclocked to 4,6ghz using a CM 212. We all know what kind of heat monster piledriver is.
Soldato
Yep, I came from a 8320 tweaked to 4.5 and it could kick out waaay more heat than the Ryzen. The air is barely warm in comparison through the rad. @alexsamam cant you jump up a res to push the GPU and maybe the 1600 will cling on for dear life some more months?
I saw mediocre reviews about CM LiquidMaster AIO series, however, I wanted to try something liquid, but it's not currently a priority. Yesterday I got my 16GB RAM, now having a total of 32, 4x8 of same model. What I noticed:
1. I can't no longer keep 2.8k as ram frequency with 4 sticks, not even 2.6k, which is surprisingly bad... currently running at 2.1k
2. With HBCC ON, CoD Modern Warfare crashes as soon as I enter the gameplay. In benchmarks, however it's not crashing. This is pretty disappointing, since I usually play this game at the moment.
About the bottleneck, I have a 27'' 144hz full hd monitor, so I can't up the resolution unless I activated VSR in radeon driver. By a few months extending, you mean the launch of next generation of ryzen cpu's? I'm not sure if they will be compatible with the old am4 boards.
1. I can't no longer keep 2.8k as ram frequency with 4 sticks, not even 2.6k, which is surprisingly bad... currently running at 2.1k
2. With HBCC ON, CoD Modern Warfare crashes as soon as I enter the gameplay. In benchmarks, however it's not crashing. This is pretty disappointing, since I usually play this game at the moment.
About the bottleneck, I have a 27'' 144hz full hd monitor, so I can't up the resolution unless I activated VSR in radeon driver. By a few months extending, you mean the launch of next generation of ryzen cpu's? I'm not sure if they will be compatible with the old am4 boards.
Soldato
Well do whatever you can to offset the load onto the GPU, like scaling and quality settings. Was trying to help not having to buy the 3600 just yet and use your current setup.
HBCC is hit or miss. It did help in situations like you've said, where there's some memory mis-management (eg Greedfall), and it was a saviour there. But sometimes it also creates problems, eg The Division 2 wouldn't start for me until I figured out it was HBCC at fault. And sometimes it just don't work properly, eg in AC Odyssey testing out 8K it would just corrupt textures or crash (as it went past the 8GB vram and into HBCC). Wouldn't upgrade in order to better use it imo.
Upgrading from a 1600 would be your best bet, though maybe at this point you might as well wait for next year's batch (which will also work on existing boards afaik). Depends if you want to wait 6 more months or not.
Associate
- Joined
- 30 Aug 2016
- Posts
- 106
5700X is roughly 20-30% faster than Vega, this isn't accounting for any undervolting/overclocks for both cards of course.
Last edited:
Today because I had some time I tried different things. As you may know, I was powering the gpu with 2 separated cables and still had problems.
1. I disconnected cable 1 and left only cable 2 (both 8pin connectors in the gpu)
2. My modular seasonic has 4 inputs x 8 pins for CPU/PIC-E cables. I disconnected the first 8 pin cable from the psu (near the mobo one) and connected there the gpu cable 2. I moved the disconnected cable to the last position.
3. I forced in bios gen3 for pcie.
Currently I have 24 hours of pc usage without any black screens, usually I had them in 1 hours or less... I still don't get way. And yes, currently there is only one cable connected to the gpu. Is there something like a faulty cable
My first install of my Gigabyte Vega 64 OC was similar. I was trying one cable, two, different ports for each cable and finally realized one of the connector ports didn't have all pins pushed down. One pin of an eight connector wasn't locked down. My system crashed several times before I figured it out. I was surprised it was actually running without full power.
Last edited:
Associate
- Joined
- 17 Dec 2018
- Posts
- 35
Hi guys
So I repasted my sapphire Vega nitro + Vega 64 with kryonaut since I wanted to run higher clocks and the African summer heat isn't condusive to that. Anyways all went well, I managed to remove and replace the 2 warranty stickers on the GPU bracket so warranty shouldn't be a problem.
Just 1 thing though. While tightening down the GPU bracket I realised that I'm screwing in more than I screwed out. I started getting paranoid about cracking the die so I just screwed all the screws untill they were of even height and left it at that.
The relief I felt when my PC booted up was unlike anything I ever experienced before with modding PC components. Anyways all temps are great and I'm running a less aggressive undervolt now resulting in higher core clocks. The delta between hotspot ,core and HBM is much lower now.
Has anyone repasted their Vega? Did you screw down the bracket all the way? Can you crack the die if mount the cooler too tightly?
So I repasted my sapphire Vega nitro + Vega 64 with kryonaut since I wanted to run higher clocks and the African summer heat isn't condusive to that. Anyways all went well, I managed to remove and replace the 2 warranty stickers on the GPU bracket so warranty shouldn't be a problem.
Just 1 thing though. While tightening down the GPU bracket I realised that I'm screwing in more than I screwed out. I started getting paranoid about cracking the die so I just screwed all the screws untill they were of even height and left it at that.
The relief I felt when my PC booted up was unlike anything I ever experienced before with modding PC components. Anyways all temps are great and I'm running a less aggressive undervolt now resulting in higher core clocks. The delta between hotspot ,core and HBM is much lower now.
Has anyone repasted their Vega? Did you screw down the bracket all the way? Can you crack the die if mount the cooler too tightly?