Exclusive: Gigabyte RX Vega56, Vega64 Gaming OC teardown

Status
Not open for further replies.
~5000 chips? Might as well not have bothered then. It's not like the reference Vega cards aren't selling, anyway.

Either way no regular gamer would want to buy in at this point. We don't just need good non-reference cooled Vega cards, we need them at a reasonable price.
 

uglyduckling81

Distinguished
Feb 24, 2011
719
0
19,060
"(particularly when AMD leaves so much room for improvement with its reference design)."
Weird thing to say when Gamers Nexus raved about how well designed the PCB is and how high quality the components are.
I believe they said it's the best PCB they ever come across. I also think they mentioned how good the Titan V was built, up to Vega standard.
 


I think it's referring to the reference cooler rather than the PCB. The cooling isn't great.
 

mihen

Honorable
Oct 11, 2017
462
53
10,890
Gamers Nexus does some things particularly well, while toms does other things particularly well. When it comes to power draw and thermals, I would go with toms.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
Read it again. It's about the cooler. The reference PCB isn't bad. It is also not perfect, but a solid layout with good component quality. AMD forgot to cool the doubler chips on the back side. This is the only improvement I would made. Thermal pads between their PCB and back plate.

 

uglyduckling81

Distinguished
Feb 24, 2011
719
0
19,060


I'm not so sure about that. They have recently upgraded their power testing equipment. For the the Titan V testing they tested the power draw of the chip manually rather than using software to do it. They have always had a good testing methodology but being fairly new they had limitations. They are slowly acquiring more equipment and getting more and more thorough.
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
Feel you so free to read all my articles about power measurement to understand, how I measure. I spent a lot of time and money in this project and developed this methodology over 5 years together with Rohde & Schwarz. And I'm comparing /sharing my data with all big AIB, they trust me. I worked 17 years in the industry and I know what I do. For power draw I'm collecting the data over ethernet, four channels for voltage and four channels for current. Calibrated with the delay of my precision clamps. I also programmed an intelligent low-pass-filter for all and I'm sure, my resultion on the scopes is high enough ;)

@all stalked out:
I'm not a newbie, benchmarks first. And I do all torture and power draw tests twice - before and after ;)

Their paste is not bad. I just tell you a secret. Most of all media samples were proofed manually from the manufacturers and are far away from retail. I know the production process in detail and can mention, if they used other thermal grease or assembled it manually. Golden samples... Not my taste. This both cards are retail. And I have also a lot of original thermal grease from the factories. Call it souvenirs :D
 

FormatC

Distinguished
Apr 4, 2011
981
1
18,990
German original:
Das sind weniger als 63%!
(less than 63%) Typical [strike]translation[/strike] correction issue. I'm sorry, but I'm not a natural speaker and I must've missed it in the 2nd check. Thx for finding, but I'm sure, it was right in my translation (just compared, it WAS right).
 
Status
Not open for further replies.