What number of AMD RX 7900 XTX’s are faulty?
I am engaged on a venture—a really dumb venture, thoughts you—and I used to be making an attempt to amass the 2 current-gen flagship GPUs: an Nvidia RTX 4090, and an AMD Radeon RX 7900 XTX.
In some bizarre stroke of luck (it has been troublesome to seek out both in inventory), I used to be in a position to get one among every this week.
(lol on the dimension distinction…)
In addition to exorbitant value gouging, Nvidia’s possession of the crown by way of GPU efficiency stays on this era, because the 4090 blows previous any competing card up to now. However AMD’s 7900 XTX was poised to be the most effective worth by way of value, efficiency, and effectivity (at the least in comparison with any Nvidia providing).
Sadly, many have experienced unexpected overheating on the 7900 XTX reference design, and der8auer even went so far as to rip open a vapor chamber to get to the bottom of the issue.
AMD initially acknowledged that 110°C junction temps have been regular, however appeared to again off that assertion as extra proof (like der8auer’s movies) got here to mild demonstrating insufficient cooling solely in a selected orientation.
After I bought my 7900 XTX within the mail, I made a decision to run some assessments. Mild gaming on my 1080p monitor did not appear to hit it too onerous, however the card was getting fairly loud because the followers ramped up.
I then ran MSI Kombustor‘s furmark donut take a look at, which mainly pushes the GPU to 100%. After lower than 7 minutes, my card’s ‘junction temperature’ hit 110°C, and I noticed the GPU clocks and energy draw drop about 5-7% because the poor followers struggled to maintain up, ramping as much as a loud 2900 rpm:
This testing was achieved with the cardboard put in in my PC within the horizontal orientation—that’s, I’ve a tower PC, and the motherboard is vertical, with the 7900 XTX put in so the followers are parallel to the bottom.
Plainly for a little bit of time, the cooler can sustain, however after 5-7 minutes, there’s simply not sufficient cooling capability on this orientation, because the vapor chamber actually runs out of steam. (Or, effectively… water, I assume.)
So I shut down my laptop, laid it on its facet (so the graphics card was within the vertical orientation), and ran the assessments once more. This time, the followers have been a quieter 1700 rpm, the max junction temperature was 92°C, and there was no throttling or energy discount for over quarter-hour:
To be full, I shut down the pc and re-tested within the vertical orientation, then once more in horizontal, and the outcomes have been equivalent. With my PC in it is typical upright configuration, I might solely get about 5 minutes of the 7900 XTX’s full efficiency earlier than it began throttling (and really loudly, at that!). If I lay my PC on its facet, I might get full efficiency all day (with about half the noise).
One thing was undoubtedly improper with the vapor chamber.
AMD mentioned “prospects experiencing this sudden limitation ought to contact AMD assist”, however when you head to the assist web page, and name the US assist cellphone quantity, it directs you to the guarantee claims web page on the web site. On that web page, it guides you thru a wizard and determines when you did not purchase the AMD card from AMD.com itself, it’s important to contact the associate producer (regardless that on this case it is the reference design, simply packaged by Sapphire).
So I contacted Sapphire assist, and bought the next:
It is a identified situation for this card, you could contact your retailer for return and buy a special model of the RX7900 sequence, Sapphire provides Pulse and Nitro sequence which don’t displays this situation.
I requested if there was any approach they may service the cardboard or substitute it, and even provide a paid improve to one among their very own associate boards (just like the aforementioned Pulse or Nitro). They responded:
we don’t deal with that sort of request right here, solely assist and guarantee service. Please proceed to test retailers website for availability.
So off to NewEgg the return goes!
I assume I will be sticking with team green this spherical.
After seeing AMD’s consultant stating:
We imagine the problem is said to the thermal resolution utilized in AMD’s reference design and is happening in a restricted variety of playing cards bought.
…why are they not pulling playing cards off the cabinets (and instructing companions to take action), in the event that they know it’s ‘restricted’ in scope? If they really know that, they need to have the ability to slim down the batch to a set of serial numbers that they may enable prospects to test.
But when they do not truly know the scope of the issue—or if its a design flaw affecting all reference fashions—they need to contemplate pulling all inventory till they’re mounted.
And possibly do not cross the buck to Sapphire, PowerColor, and different companions who’ve to provide their prospects a foul expertise, since no comparable alternative playing cards might be had with out paying scalpers on eBay.