1 of 1 people found this helpful
Put the old RAM back in and check it.
1 of 1 people found this helpful
It sounds like it's been damaged in transit., but that's not exactly something that can be diagnosed easily over-the-internet.
Make sure that you've Seated it properly (take it out, and put it back in firmly) … Make sure you've connected both PCI-E Power Connectors., if you have Multiple PCI-E Power Connectors, try different combinations of the available connectors.
Try running with the Original 8GB System RAM, then the New 8GB System RAM... it's possible that one or more of the new RAM Modules is faulty., and yes even if it's just one of them, it will prevent the ENTIRE System from Completing POST, normally with a 1 Short, and 2 Long Bleeps via the PC Speaker., but people tend not to have them connected anymore.
Failing this, borrow or acquire a cheap'n'cheerful GPU that you can use as Primary and put the Vega in the Secondary Slot., just to see if the System even recognises it's there. It's possible the V-BIOS has become corrupted and will need to be Re-Flashed., unlikely but possible. And again it's a similar issue to "Bad RAM" as the Card will simply not POST., but this is more difficult to diagnose and work with for Graphics Cards.
Beyond that., you can either RMA it (assuming it's still within Warranty., it should be) or take it into an Electronics Repair Store., like iFix or such.
It's going to either be a very simplistic issue, or the GPU SoC itself has been damaged and thus well dead card; but good luck getting the Postal Service to take responsibility for that.
1 of 1 people found this helpful
After checking the RAM since that was also changed, it still may be a PSU issue. If you are able to run MEMTEST86 I would do that just to eliminate bad RAM Module.
According to AMD Specs you need a minimum PSU of 750 watts : Radeon™ RX Vega⁶⁴ Graphics | AMD .
According to this Website that post various PSU minimum requirements for most AMD & nVidia GPU cards mentions for Air-cooled Vega64 you need a minimum of PSU of 700 Watts (total system use) and for Liquid-Cooled Vega64 you need a minimum PSU of 900 Watts.PSU REQUIREMENTS - RealHardTechX
But even with a slightly underpowered PSU, the Vega64 should have started up and worked.
Make sure you have all GPU Power connectors connected directly from the PSU to your Vega64. Also make sure you connected all Motherboard Power connectors including to the CPU AUX power connectors.
EDIT: Best way to check to see if your Vega64 is defective is by installing it on another computer. If it has the exact symptoms, your card is probably bad. if it starts working on another computer then it probably indicates something is wrong with your computer setup.
So I tried to do what you suggested in your edit and connected it to my old pc which has an msi g31tm-p21 motherboard everything running on the new PSU. The card fan spinned up like its nothing so I guess that's fine although I couldn't get the signal from the display port on the backside of the card. I will most certainly try with the old RAM as suggested by leyvin.
I should mention a few things just to make things more clear. I have a GA-AX370M-DS3H motherboard which supports the 2400G amd processor with integrated vega 11 graphics. I'm not sure but don't want to rule out this option that this might cause some issues too. I haven't addressed it yet (haven't tried disabling it). To add I've had no issues turning the PC on and getting the signal through the card beforehand.
What I've tried is switching the 8-pin cables up but that didn't work. I reinstalled windows and deleted all files so that removed any drivers that we're present. This didn't help so I decided to try and flash the BIOS to an older version which also didn't work. Before all of that I also tried powering the system on without the card, HDD, with only 1 stick of RAM in different slots and combining all of it in different ways. Therefore I think the PSU is ok.
I'll try some other stuff and report back.
3 of 3 people found this helpful
So am I correct in assuming from your post, that you are using a 2400G? When you get no signal from the back of the GPU, do you in fact get signal when you move the cable to the ports on the motherboard? If so, your PC is running with the IGP setting for the primary display adapter. You should be able to change that to PCIE in the UEFI. If that still doesn't work, try switching the actual PCI-E slot the Vega GPU is installed in.
That's right. I have the ryzen 2400g. Good thinking. As I stated before with the integrated graphics I went on to check my default display setting in my BIOS. Sadly it was already set to PCIe 1 so thats that.
Other things that I've tried is running the system with the old RAM. I even tried to connect it to my friends rog strix b350-f with a ryzen 1600. This had the same result with the card just lighting up but no fan start at start up. This was rwally dissapointing. The PSU was my 700W one. He has a 500W PSU so we didn't even consider using his.
My only hope is still the fact that my PSU is not good enough because of the fact that I tried to check if the card is working by connecting it to my old motherboard with an intel core 2 duo. The mobo had a 4 pin connector for the processor. I'm going to a store where I'll check out on some of their test benches if the card runs or not. Wish me luck.
1 of 1 people found this helpful
A lack of Power won't result in the System (Card) not Powering Up., especially given when not underload the Card will be drawing between 13 - 48w.
It's only underload will it peak at it's Maximum (Momentarily) before settling into it's Avg. Draw for the Application.
At this point what will happen during that Temporary Max Draw, is it will trigger the Power Surge Trip (i.e. More Power is being Requested than can be Supplied) … Advanced Power Control Interface will freak out and shutdown the System., almost identical to as if a Component trips the Temperature Sensor.. as it can even result in the need to switch off (or remove the Socket Plug) in order to reset the PSU to turn back on.
Keep in mind that the PSU can't tell what direction a Surge comes from (i.e. Socket, Motherboard or Device) it just knows one happened, so it's trying to be helpful and save your hardware.
Now as best as I can tell., there is obviously an issue with the Card itself (i.e. it has been damaged in transit)… the Fans not spinning up, well that can be caused quite literally by never reaching the Fan Stage of the BIOS POST; which could potentially mean that the GPU has become unseated (and requires re-balling) … an issue that wouldn't be possible if AMD would switch to Socketed instead of Embedded GPU Designs, this remains an absolute bewilderment to me given since the introduction of GCN these are "Complete" SoC, meaning for all-intended-purposes the Card itself is little more than providing VRM (Power), BIOS, VRAM and Display Links... all of which are basically identical.
As the (Solder) Ball Grid Array is the source of many board failures, especially as said connections can melt and warp over-time due to lack of airflow or such; well providing a Rigid Socket with Pins (either ZIP-PGA like Ryzen or LGA like Threadripper; doesn't matter) it in essence entirely resolves said seriously common issue.
Plus it would given how said GPU SoC are for the most part interchangeable, could allow AMD to sell them separately either as Replacements or Upgrades.
Like it made sense years ago for Graphics Cards to be Complete Solutions as they were comprised of Multiple Chips that were unique from Generation-to-Generation or even within the same Budget-Hyper Performance Processors (as they used to segment functionality and features)… but today it makes no sense for what., an extra $15 (Total) Production Cost?
But eh, w/e point here is... this is your likely culprit; which can either be done professionally quite easily (but is time consuming, as it takes about 45-80min in total not including creating the BGA Stencil; which can be a pain given these aren't typically provided "After Market", so they tend to need to be manually etched; this can take another hour or so. (Again another reason why Sockets would be better)
Yet, it could also be that a Micro-Transistor has been dislodged / fallen off / shorted out. In the past a common issue with Graphics Card Failures, well it was either DRAM (VRAM) developing a fault or one of the thousands of Micro-Transistors basically committed seppuku; and by the gods is that a pain in the rear to track down. As you essentially have to visually check over the board., if you can't find what's missing, run through the various Traces to discover the line that has a fault... it's just time consuming as all hell.
Had to do it recently to resurrect my Intel Pentium II that just randomly died., both the Processor Board (as they're on their own boards) as well as the LX440 Motherboard; which was a "Fun" Weekend... to be honest, if it wasn't a sentimental system., I'd have just thrown the parts in a box for "Salvage" in later projects and bought new from Electromyne.
What I'd suggest is RMA (Return to Manufacturer) and get them to refurbish / replace., honestly be the cheapest and easiest approach; and unless you Pre-Ordered and received it at Launch, well you should still be within Warranty; which typically only doesn't cover you if the damage was done via Overclocking … heck even then some will still actually be okay providing it, as well "Good Will" usually means next purchase will also be from them., so it pays forward for them to seem like they go above and beyond.
It's weird how some Companies don't understand this aspect of Customer Service, but eh... their loss.
Ok so I have finally got the card checked out and it's fine.
Its the RAM thats causing the issue. I can't boot at all with the old one. The new one is ok. But I can't boot with the graphics card connected. Only on the integrated. This is peculiar. I checked in the BIOS for the default video output and it is set to PCIe. Not sure what I should do next.