Is AMD Vega's Package Construction A Problem? A Closer Look

Status
Not open for further replies.

NP

Distinguished
Jan 8, 2015
74
15
18,535
Q: "How relevant is this, anyway?"
A: "Not relevant, just something random to say in the futile effort to take away the pain from the flopped vega. It looks like a duck, talks like a duck, and walks like a duck. Guess what? It's because it is a duck! Air cooling not enough, but even with huge power draw water cooling wont translate to results, neither in the sense of performance or value. Any TIM spin - or messianic faith in Gigabyte's propeller gadget over AMD's - is just sad. Oh well, better product next time."
 
Ahh please...

You are buying yourself problems... died for a second time in less than 10 months! You are buying yourself problems. Worst card I ever bought!

This review is from: EVGA GeForce GTX 1080 FTW GAMING ACX 3.0, 08G-P4-6286-KR, 8GB GDDR5X, RGB LED, 10CM FAN, 10 Power Phases, Double BIOS, DX12 OSD Support (PXOC)

Pros: -Best Price/Perfomance ratio for any 1080 GTX
-OC easily over 2000 MHz out of the box
-No more cooling required for hitting the 2100 MHz wall of the 1080 GTX
-Incredible customer support from EVGA, 5 stars easily! They paid for both shipping way TWICE with a little convincing.

Cons: -Reliability and Quality Assurance is not there... Don't sugar coat this, it's a bad PCB with many weaknesses
-Overheating of VRAM problems making the card degrade over time until it dies
-A brand new EVGA SuperNOVA power supply blew up with this fantastic 1 month old card
-Nvidia drivers not stable at 4k in windows 10
-Closing or opening a game can make your system crash
-Not as powerful as I was expecting, should have sticked to my dual 290x or waited for a 1080 TI
-Random Displayport and HDMI signal drops when the resolution change in games settings... it's crazy, it's that bad. I cannot say if this is Nvidia drivers or the card, however my IGP never shown any issue with the same cables

Other Thoughts: Update:
My card died for a second time in less than a year. The last time I saw something like that was with a Gigabyte GA-965P. the board died 3 time in 16 months. This is unacceptable and you end up RMA the thing until your warranty is valid. After that, you are on your own. Problems with these cards are going to appear more and more with the time passing as the weak VRAM degrade due to overheating issues and the lack of thermal pad.

I had enough, I ordered a new MSI 1080 TI and I am going to sell my RMA card to an unlucky guy. I am not going to wait for another 3 weeks without a desktop waiting for the replacement.

I learned that my credit card could have covered the cost of the card because newegg refused to take the card back after 30 days. Inform yourself, some cover you up to 90 days. Of course, newegg didn't informed me.

Original:
I bought this card because it was 2 slots only and was providing enough cooling for hitting the 2100 MHz barrier of the 1080 GTX. Buying anything with more cooling power is a waste of time and money since no one can push these cards further. This was the perfect choice.

I woke up one morning and my desktop was shut down. I tried to turn it on but didn't succeed. I suspected a Power Supply failure, however it was the graphic card that died. The card died in idle during the night. Anyway, I was building myself a new mini-itx computer and I decided to test my system with my graphic card. Big mistake, the card exploded and the power supply in my new system was an EVGA SuperNOVA B2 750W. Hard for EVGA to dodge that one. My mobo died also and was perfectly working.

Lessons learned, if your PC doesn't turn on, remove the GPU and test it again.

Also, the new card is showing the same display feed dropping behaviors on a fresh install. It is really annoying and make windows 10 literally crash. In game crashes are rare for me, but jumping in or out of a game is problematic and can make my PC crash due to resolutions conflict issues that are OS or driver related. For those praising Nvidia drivers, I can tell you I never noticed something similar with my 290x AMD cards in crossfire. My system was more stable with my Powercolors CF.

I cannot recommend this product at all. It's an important investment and I am left with a bad experience, I had to RMA from 3 different companies. However, EVGA paid for all the shipping charges even if I am canadian and I received my replacements in less than 3 weeks.

1 egg for product
2 egg for Newegg support (didn't provide any help even if barely 45 days of ownership)
5 egg for EVGA customer support (at least you know they are going to take care of you if something happens)
 

techac

Prominent
Aug 22, 2017
3
0
510
Seems like you are trying to come with issues that don't exist yet or that there are no complains about.. AMD is selling huge amounts of professional and gaming cards and haven't hear any issues..
 

therealduckofdeath

Honorable
May 10, 2012
783
0
11,160
I think, a lot of GPU manufacturers automatically voids your warranty if you take off the cooler, so, is this really a big problem? Also, the wording, implying that it's a general problem for anyone buying, is a bit [forbidden word beginning with t], isn't it?
 

knowom

Distinguished
Jan 28, 2006
782
0
18,990
a lot of ram doesn't even use heatsink's at all though this is HBM2 and GPU VRAM we are talking about so it's slightly, but even GPU's for the longest while didn't bother using them so take it with a grain of salt I guess on it's overall importance
 

robertcainphoto

Prominent
Aug 22, 2017
1
0
510
.1 mm equals .0039 inch, close to the diameter of a strand of human hair .
Can't imagine this causing a big problem with thermal heat dissipation.
 
10TACLE,
Huh? First of all, the Asus Strix will beat the reference VEGA64 air-cooled card. You also linked a page showing VEGA64 lose to the GTX1080 in one game and beat it in two others (DOOM and W3).

*wrong info in Witcher 3 BTW. That GTX1080 FE card next to the 1080Ti is obviously wrong.

I don't recommend VEGA64 though.
 
REDGARL,
(not sure if you mean to reply here but.. )
I have a GTX1080 FTW3, W10, 4K monitor and have had none of the issues you describe. Also using an EVGA G2 750W power supply.

You mention "overheating" and lack of a thermal pad however EVGA fixed that. You could return the card (which I guess you did due to failure) and got a replacement with the thermal pad applied.

No major issues with many, many games for almost a YEAR that I've had this card.

I suspect your EVGA PSU died and took the card with it. There's no good reason i can think of that a card dies and takes a PSU with it. That begs the question why the 2nd card died... bad luck or possibly some problems with local power surges?

*May want to consider the APC Line-R 1200VA. It's an AC line conditioner.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
It is only a translation, but please read it again.

The second thing: the HBM2 is hotter than the GPU. In some cases my HBM2 on the Vega64 was running significant above 90°C and to be honest - this is nothing to have it in such a system. Four screws, four corners? To have the perfect contact in the complete surface, you need to fix it totally plain. If the whole surface is not plain, you need in each case stoppers, to get in each corner the same pressure.

If a cooler was constructed for molded packages and is now used for unmolded, the pressure might be too high and the interposer will simply crack. If the pressure is in the opposite case too low, you will lose a lot of performance. I talked with a few AIB and the whole industry is more or less in panic mode. A few R&D guys bricked their card only with too much pressure and this alone is a good indicator for an issue. You must work with stoppers for the screws but with a different hight you won't get the optimum.

@redgarl:
I tested this VGA card here and punished EVGA for their design. I suggested them a pad mod and a modified BIOS. EVGA was here in my lab and we proofed the pad mod thing before they used it in the mass production for the last ACX cards. All this changes in the current iCX design are more or less based on my analysis, that I send them in August 2016.

I'm working not only as editor and I have many products and prototypes here, to test the early samples before the MP starts. This is the reason why I have so good contacts to the factories and R&D guys of a lot of AIB. And I know, that this package chaos is not only a small thing, it is a problem for each vendor.

 

nirvansiga1

Prominent
Aug 23, 2017
1
0
510
when polaris first launched they showed two rx 480's in crossfire matching a 1080.
Now, iirc the x-fire setup would draw around the same amount of power (~300 watts) as vega and be just as fast.
given vega's performance and tdp they could have saved a huge ton of cash on r&d by just releasing a dual-gpu 480/580.
anyone else wondering the same?
 
We need more power!!!
https://static1.squarespace.com/static/51e3f76ee4b07f69602a6fcc/t/53177c63e4b0ab76f8ffae34/1394048115517/More+Power.png

Captain I'm giving all she's got to stay within 250 watt tdp.

Scotty increase the tdp to 400 WATTS!!!
http://www.tomshardware.com/reviews/radeon-rx-vega-64-water-cooling,5177-3.html

Captain we still aren't making frames fast enough.

Scotty solder another card on there, downclock them both by 20%, increase the tdp to a combined 600 watts and call it the Vega 64x2 Most Awesomely Water-Cooled Edition

No link for that, but I wouldn't be surprised if it did come in a similar form.
 

jleppard

Reputable
Jul 28, 2015
4
0
4,510
Funny how a small company EK can make a water block that fits on all this so called 3 different HBM packages with no problem and keep it a 40 degrees under full stress loads without all the NEW scare.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
I talked with Atilla from EK and I'm sure, they never tested it with an unmolded package that is 0,1mm higher. We had longer talks with other cooler vendors and ODM, nobody got this foils and info so far. If you try to mount an optimized water block for a molded package on a PCB with an unmolded package and the viscosity of your TIM is too high, the interposer may crack. I have no idea, why AMD has not shared this foils earlier...

The BOM (bills of materials) for all AIB was made in May. But in May nobody got this info. A lot if issues are simply home-made by AMD. This is one of this. Too bad.
 
Photonboy, I posted my review from newegg because toms was talking about design issue. Well, EVGA released a piece of junk. They are on my blacklist. I am having a platinum 1000 W Newton R3 from Fractal Design, it's a higher end power supply. Both card died with it, however my 290xes never had an issue for 3 years.

I switched my 1080 FTW for an MSI 1080 TI Duke and none of the issues I was having before are there anymore... especially the display signal drop problem. Digital signal to the displayport and hdmi ports is inadequate.

http://www.tomshardware.com/news/evga-addresses-geforce-1080-temperature-problems,32967.html

It is a known issue. These cards are horribly designed. They made 2 revision of the card for a damn reason, it is freaking lemon.
 


Yeah those 1080 Ti figures are way off. Not even sure why they put them in there. But I think you misunderstood the point of that link. The point was a performance increase comparison of the reference RX 64 to the ASUS Strix AIB variant with supposed better cooling and higher clocking. If that link is valid, it's far below the jump in performance of a reference 1080 to an AIB partner one with better cooling and higher clocking. It was not meant to be a direct comparison to the 1080.

 

TJ Hooker

Titan
Ambassador

I still don't see what your anecdote has to do with this article. OK, maybe one of EVGA's cooling solutions was sub-par (although given that the card has an average 5 star rating on Newegg with 350 reviews, it appears the vast majority of people didn't have a problem). Even if that's the case, it's an issue with one specific cooler from one company. Nothing to do with the overall manufacturability of Nvidia graphics cards.
 
"You are buying yourself problems... died for a second time in less than 10 months! You are buying yourself problems. Worst card I ever bought!

This review is from: EVGA GeForce GTX 1080 FTW GAMING ACX 3.0, 08G-P4-6286-KR, 8GB GDDR5X, RGB LED, 10CM FAN, 10 Power Phases, Double BIOS, DX12 OSD Support (PXOC)"

The EVGA SC series has been plagued with issues since the 500 series where fried VRMS were the major concern when over clocking. On the 970, 1/3 of the heat pipes "missed" the GPU. The 10xx card's VRMs also were going up in flames.

The term AIB (Add-in Board) should not apply to the SC series as historically, the PCB has nothing "added in" . It's a reference board with a better cooler. One of its more glaring weaknesses has been the inadequate VRM and Memory cooling where EVGA decided to try and save 11 cents by not including a thermal pad of other means as is typical for other AIB board partners.

With the 10xx nerfed by Boost 3 and the performance gap between the FTW and SC series shrunken, EVGA figured they could go cheap on the FTW series too and again omitted the thermal pads resulting in all the burnt PCBs which caused the recall / hardware fix kit.

I'm not sure how meeting their legal obligations qualifies and picking up your shipping (essentially bribing their customer not to make a fusss and trash the company all over the web) as superior tech support. Our last experience with EVGA for a FTW card that didn't meet advertised specs out of the box, was 20 support calls, 18 months, 5 RMAs, the last being that they sent a next reference generation card. But the most frustrating part was that on the 2nd thru 18th call, they made you go thru the same 30-45 minute test procedure so that you could prove for the upteenth time that other components were not at fault.
 
So far I've been extremely happy with my EVGA 1080 Ti SC2 Gaming (iCX). It is a killer overclocker that rivals my former GTX 970 ACX 2.0+ GPUs in overclock performance % improvement. I've been buying EVGA solely for eight years, starting with the GTX 275. I hadn't had any problems except for the first time ever had to RMA a GTX 970 with a failing DVI-D port. It was a painless operation and they cross-shipped so my wait was minimal (not every GPU vendor cross ships).

With that said, I'm starting to get a little concerned that EVGA is being controlled more and more by bean counters who care nothing about R&D investment. The recent G3 series PSU problems that Tom's has reported is further troubling. I'm keeping an eye on my GPU closely while gaming. But it is nice to know you have top tier customer support should you need it. And EVGA is not the only AIB partner out there that people are reporting issues with on their GTX 108x GPUs.

Just this morning I came across a guy here reporting major issues with his Gigabyte GTX 1080 G1 Gaming card he had just bought. A quick check on Amazon reviews of that card showed nearly 20% of buyers reporting major issues just like he did: intolerable coil noise, outrageous temps at 80C+ even at 100% fan speed, and just flat dying being the three most reported issues. That's a lot of people reporting the same problem. I have not heard any major problems between MSI and ASUS variants of those series cards.
 

RomeoReject

Reputable
Jan 4, 2015
239
0
4,680
Is there any reason why they couldn't have just used one manufacturer for each card, rather than going half and half on both cards? Would've solved much of the issues listed here.
 
Status
Not open for further replies.