7900XTX Hotspot Temperature Issue tested (der8auer)

[Kenrou]

Remember having the same issue with my XFX 5700xt, hotspot was 30c+ hotter than the core, a remount with NT-H1 and Gelid pads lowered it to 10c-15c difference (depending on voltage/watts used at the time), mounting pressure felt the same

 

[SF101]

They also added voltage locking to fix these "issues" this brings me flashbacks from the 2019 too 2020 drivers on the 5700XT ..
My Power-color Liquid Devil was super gimped from that day forth with those locked voltages etc and could not be bio's flashed.
This would not make my happy at all id be so pissed if this was me.
I damn near swore off AMD Gpu's over it.

 

[Kenrou]

After further testing, Der8auer claims it's an issue with the vapor chamber on SOME of the stock models, RMA and the partner AIBs that use the stock cooler will likely end you with the same problem, so.... Wait and see what AMD does?

Hell of a way to start the new year...

 

[Soulcatcher668]

Watched this morning.... not looking good for AMD. We will see how they handle it.

 

[Janus67]

Came to post the same video. Definitely not good news.

 

[SF101]

Damn AMD dropped the football .. a freaking vapor chamber failure? this is a ridiculous fail.
Nothing to do with the card itself just a huge fail on the cooling side .. how can this pass testing ever?

 

[Woomack]

After further testing der8auer says the same as multiple other websites It's all over the web for a couple of days ... or I'm so lucky to get spammed by twitter news from various sources.
It's related only to the reference design. Custom cards don't have problems with that. If it repeats like in the last gens then reference cards will be barely available anyway.

 

[Kenrou]

Not enough coolant in the vapor chamber ?

 

[Soulcatcher668]

Not enough coolant in the vapor chamber ?

So it looks like somebody missed a decimal point on the fill calculation? Oof.

 

[EarthDog]

World seems like it's falling, but.........who, HERE, buys reference cards anyway?

I mean, yikes... but can someone give me some cliff's notes? Is this only in a certain orientation? And not all cards are affected? I'll have the time to watch all of these on the plane tomorrow, lol.

 

[Woomack]

I wonder if it's a matter of a specific batch or just a design flaw which is repeated in all reference cards. There are not so many reference cards around and barely anyone is even sending them for reviews. Only fanboys buy them on the first day after the premiere, when other cards are not available.

 

[EarthDog]

That's my point. Though, at launch, I think many of the 'big' reviewers (read, not us apparently) get their samples from AMD which is reference. To that, I wonder, was it never caught???

I still have no idea what orientation this happens or situation as I don't have the time to watch the vids rn.

 

[Woomack]

Some, if not most, reviewers aren't really good at what they do and/or don't want to say anything bad about vendors that deliver them "free" stuff.

 

[Kenrou]

I still have no idea what orientation this happens or situation as I don't have the time to watch the vids rn.

Problem is the hotspot, core never seems to go above 70c. He mounted them in different positions and even "tipped" them while stress testing, vertical seems ok-ish until tipped, then goes 110c in a couple of minutes, horizontal is hot from the start.

 

[EarthDog]

So vertical is the way to go. I see. Too bad horizontal is the way most cases mount graphics.

 

[Kenrou]

One of the samples had high temps regardless of orientation though (was actually worse vertical), also expect it to happen "in the wild"

 

[mackerel]

After further testing der8auer says the same as multiple other websites It's all over the web for a couple of days ... or I'm so lucky to get spammed by twitter news from various sources.

That video is from a couple of days ago and likely the reference of many of those other sites. Anyone can speculate as to the cause, but it has more meaning when there is testing done as to why it might be happening.

A follow up video was released today where he opened one up. Doesn't really get us any further but I found it interesting to poke inside.

World seems like it's falling, but.........who, HERE, buys reference cards anyway?

The last one I got was the Vega 56 so it was a long time ago. However that was during the onset of the first mining boom so any GPU was a good GPU then. I did have the 1070 FE and 980 Ti FE and they were fine for the time.

Today, I wouldn't mind the 30/40 series FE, if for no other reason than they tend to be cheaper than many AIB cards and aren't bad at all.

So vertical is the way to go. I see. Too bad horizontal is the way most cases mount graphics.

The interesting part of the problem is if you take a vertically working GPU, tilt it horizontal so it then misbehaves, it doesn't restore performance by rotating back to vertical again. Presumably if you reduce the load and let it reset it would be ok again. So yes, this does seem to be a cooler problem. Is it a bad batch or more fundamental design flaw? Still at least it seems the problem is on the cooler, and the rest of the hardware is fine as far as we know.

 

[Kenrou]

For those that don't have an issue with vapor chamber but still have high hotspot temperatures, a solution seems to be a better DisplayPort cable? Some dell (and other cheap quality brands) 20pin cables seem to send different amounts of voltage than needed?

 

[EarthDog]

Well no ****......... Weird.........above my head on how that works/why it's happening.

Will watch Jesus vid on the plane, lol.

 

[Johan45]

Well no ****......... Weird.........above my head on how that works/why it's happening.

Will watch Jesus vid on the plane, lol.

Video Jesus doesn't have a clue what's happening either. He was relaying user comments and experiences