AMD Vega Hardware Reviews
- Thread starter ArkeoTP
- Start date
You know that tools like GPU-Z are having huge problems to read the power draw correctly for Vega?
Toms has done some interesting testing with a water cooled RX56 to eliminate the cooling limitation and even undervolted to 1V power draw remained the same. The only found some advantage at max. Powertarget.
http://www.tomshardware.de/vega-und...aufnahme-messungen,testberichte-242393-2.html
Toms has done some interesting testing with a water cooled RX56 to eliminate the cooling limitation and even undervolted to 1V power draw remained the same. The only found some advantage at max. Powertarget.
http://www.tomshardware.de/vega-und...aufnahme-messungen,testberichte-242393-2.html
Last edited:
I've got 5 bucks on the ROPs...
Why the ROPs?
But doesn't it light up even under pure compute loads? ROPs would be inactive then wouldn't they?
I'm thinking maybe my 5 bucks would go to some kind of central data switch/crossbar, lots of wires crossing, lots of bits passing back and forth. Lots of heat. *shrug*
Is there any way to test how hot your hotspot gets under a workload that's basically data transfer?
Could HBM clocks be a reflection of some kind of power/heat issue for the soc?
Well, considering they share the same heatsink hotplate (or vapor chamber, in the radial fan reference design), some of the vega ASIC heat would leak over to the HBM side... Or am I misunderstanding you?
Well, considering they share the same heatsink hotplate (or vapor chamber, in the radial fan reference design), some of the vega ASIC heat would leak over to the HBM side... Or am I misunderstanding you?
I was wondering if something in the SoC - perhaps the memory controller(s?), ran particularly hot when HBM clocks were high and the bus was being stressed. Vega has a stupid amount of bandwidth, and the traffic to and from the memory has to be co-ordinated at a centralised location even if the interface is spread out around the edges of the chip. I was just wondering if there was some link between Vega's hotspot related throttling woes and the way Vega was handling data transfers... *shrug*
gamervivek
Regular
Tested it with 700Mhz and 900Mhz on memory, no difference. Extra fans don't change anything either.
Well, there goes that idea!
D
Deleted member 2197
Guest
ASUS Radeon ROG RX Vega 64 STRIX 8GB review - Introduction
http://www.guru3d.com/articles-pages/asus-radeon-rog-rx-vega-64-strix-8gb-review,1.html
http://www.guru3d.com/articles-pages/asus-radeon-rog-rx-vega-64-strix-8gb-review,1.html
Well that Italian website which name I can't remember said that Asus gave up on the BIOS because the driver was the problem. Still strange that asus ask to take off a review because they chip the wrong bios? so they make a mistake thinking that it was a bios bug or they though they could modify the bios but ended up realizing it wasn't a solution? I mean you dont ask to take down a review unless you are pretty sure of what you are saying.
D
Deleted member 2197
Guest
First custom design: ASUS ROG Strix Radeon Vega 64 OC Edition tested
https://www.hardwareluxx.de/index.p...-strix-radeon-vega-64-oc-edition-im-test.html
https://www.hardwareluxx.de/index.p...-strix-radeon-vega-64-oc-edition-im-test.html
My German is lousy, but I can sort of make out some of the text...
Interesting that they actually compare the regular Strix edition and Strix OC head to head in the overclocking section, so you can see what difference those last two extra letters make (about 5 FPS give or take, by the looks of it, at a cost of an extra 50W power draw...)
Also, their card wouldn't hit 1100 on the HBM (and it's the un-epoxied version of the ASIC), but I guess that's just my bias showing...
itsmydamnation
Veteran
Keep shifting those goal posts Grall or do I have to quote what was said
D
Deleted member 2197
Guest
I don't even know what the funk you're on about. Like it's my fault they couldn't hit 1100? What? Lol.Keep shifting those goal posts Grall or do I have to quote what was said
Get effin over yourself already.
...Anyhow, the Guru3D review is very very curious. In quite a few benches the bog standard Vega is actually faster than the ASUS Strix (or ASUS card is only barely faster - literally by 1 FPS or somesuch), which really should be impossible unless something is bonkers wrong somewhere - like, their review card is a lender which has made the rounds on other sites already and been ripped apart one too many times and just wasn't put back together properly again or something. Or some driver confusion maybe, I dunno what's going on there.
Last edited by a moderator:
Rootax
Veteran
Jayztwocent saw the same thing, the vanilla amd card clocks a little higher.
I guess the asus cooler is not designed correctly for Vega (bad gpu/hbm contact, bad vrm&co contact, ... ?)
But, I get the need for custom cooling, but why did asus made a custom vrm/power delivery design too ? Amd stock is top notch, hell it's one thing in Vega which is nearly perfect, and that's the thing they change... ?
I guess the asus cooler is not designed correctly for Vega (bad gpu/hbm contact, bad vrm&co contact, ... ?)
But, I get the need for custom cooling, but why did asus made a custom vrm/power delivery design too ? Amd stock is top notch, hell it's one thing in Vega which is nearly perfect, and that's the thing they change... ?
D
Deleted member 13524
Guest
Maybe to save money?
Similar threads
