MSI: “Damn, [RX Vega] Needs a Lot of Power”

There is one plus from high power consumption, being that it's a massive turn-off for miners.
It'll still be under produced and under-supplied by AMD, and price gouged by retailers. Forget MSRP for the first few weeks or months.
 
The Instinct get 24.6 TF at 300W running at 1500. I am just wondering what is going to happen with the 375W one. We might be very surprised... The Instinct should also be clocked lower than Vega RX. So this could be very interesting. Maybe the Watts are actually doing something.
 
RX Vega would have to be twice as fast than Fiji to reach 1080ti speeds. Yet its only clocked ~50% higher and actually having slightly less bandwidth.

https://www.computerbase.de/thema/grafikkarte/rangliste/#diagramm-performancerating-1920-1080
upload_2017-6-11_12-58-52-png.27428

upload_2017-6-11_12-59-30-png.27429

upload_2017-6-11_12-59-51-png.27430
 
Ye, computerbase.de tends to be very pro AMD. So you are right. Cherrypicked AMD site.

I've never even heard of that website.

Also we are supposed to take someones' word that reposted what he says is a translated quote as anything more than rumor. He isn't even credible enough for MSI or AMD to reply back to him reading his forum posts on the link. While I'm not saying it isn't true or not true, if it is, why on earth did Intel decide to use Vega embedded chips on some of their motherboards for their latest CPU's if Vega is this bad on power usage. Just tossing out my 2 cents.
 
I've never even heard of that website.

Also we are supposed to take someones' word that reposted what he says is a translated quote as anything more than rumor. He isn't even credible enough for MSI or AMD to reply back to him reading his forum posts on the link. While I'm not saying it isn't true or not true, if it is, why on earth did Intel decide to use Vega embedded chips on some of their motherboards for their latest CPU's if Vega is this bad on power usage. Just tossing out my 2 cents.

The quote came from MSI and its translated correctly.

And did Intel? I dont think so.
 
Ye, computerbase.de tends to be very pro AMD. So you are right. Cherrypicked AMD site.
Look at just a few cards on that chart.

The FuryX trading blows with a 480/1060? Granted a rare game here and there maybe, but given many a bench of newer games the FuryX can at times approach 1070, and add to that the higher resolutions the FuryX and most AMD card tend to outperform their Nvidia counterparts. However your crappy chart shows almost the same level across all resolutions. So BS claims can be made against that site and chart with relative reason.
 
Look at just a few cards on that chart.

The FuryX trading blows with a 480/1060? Granted a rare game here and there maybe, but given many a bench of newer games the FuryX can at times approach 1070, and add to that the higher resolutions the FuryX and most AMD card tend to outperform their Nvidia counterparts. However your crappy chart shows almost the same level across all resolutions. So BS claims can be made against that site and chart with relative reason.

Maybe because the Fury X doesn´t do so well as some proclaims.
The list of games tested is here:
https://www.computerbase.de/2017-05/grafikkarten-testsystem-2017/

If you want a few more AMD sponsored titles in you can look here and avoid any aftermarket cards as such including updated cards like 500 series, 9 and 11Ghz cards 1060/1080 cards.
https://www.techpowerup.com/reviews/MSI/GTX_1080_Ti_Gaming_X/30.html

perfrel_1920_1080.png

perfrel_2560_1440.png

perfrel_3840_2160.png
 
Maybe because the Fury X doesn´t do so well as some proclaims.
The list of games tested is here:
https://www.computerbase.de/2017-05/grafikkarten-testsystem-2017/

If you want a few more AMD sponsored titles in you can look here and avoid any aftermarket cards as such including updated cards like 500 series, 9 and 11Ghz cards 1060/1080 cards.
https://www.techpowerup.com/reviews/MSI/GTX_1080_Ti_Gaming_X/30.html

perfrel_1920_1080.png

perfrel_2560_1440.png

perfrel_3840_2160.png
See. That second list is closer to what we normally see. Hence the distrust of your first list. They have something terribly wrong with how they test and for good reason why most will not give it an ounce of respect.
 
See. That second list is closer to what we normally see. Hence the distrust of your first list. They have something terribly wrong with how they test and for good reason why most will not give it an ounce of respect.

Normally see? You mean it depends on the list of games? You better be happy Dirt 4 isn't included then.
 
Last edited:
Normally see? You mean it depends on the list of games?
no i mean in a game by game review, as in typically. As I mentioned previously, the 480/1060 has in a RARE game or 2 shown results close to the FuryX, whereas TYPICALLY the FuryX is closer to the 1070. Your first chart would need to be trying to get that result, as in biasing the result. Also look at all AMD cards in their list, resolution does change the ranking nor show much of a change in performance %, where nearly most individual game reviews do. Again their results seem more akin to outliers than a general expectation.
 
no i mean in a game by game review, as in typically. As I mentioned previously, the 480/1060 has in a RARE game or 2 shown results close to the FuryX, whereas TYPICALLY the FuryX is closer to the 1070. Your first chart would need to be trying to get that result, as in biasing the result. Also look at all AMD cards in their list, resolution does change the ranking nor show much of a change in performance %, where nearly most individual game reviews do. Again their results seem more akin to outliers than a general expectation.

It depends a lot on the per game doesn't it.

Dirt 4? Conan Exiles, Ark and so on.

But I am sure its better to include some failed AAA sponsored game with abnormal performance ;)
 
Lets be honest. There is no reason for a fast card. If you are playing CS GO at 400 FPS, you're an asshole anyway.
Or hey, maybe there are other games that use videocards besides one with small, contained maps running on a 13 year old engine.
 
Green has been killing in performance per Watt since Maxwell compared to Red. I love my AMD but I build almost all SFF boxes now and the power/heat situation just favors nvidia. I still am willing to pick up a 1070 tier vega card if I can run it with my sf450 corsair psu and Ryzen 1700. If not I KNOW a gtx 1080 will drop right in. Not long to see how it all shakes out.
 
Lets be honest. There is no reason for a fast card. If you are playing CS GO at 400 FPS, you're an asshole anyway.
Or hey, maybe there are other games that use videocards besides one with small, contained maps running on a 13 year old engine.

Not if you want to play CSGO in 4k THUMPer. How am I an asshole if I want 122fps at that res?
 
Not if you want to play CSGO in 4k THUMPer. How am I an asshole if I want 122fps at that res?
I'm not arguing against you, that's my point. CS:GO is one of the better performing games out there. There are plenty of games that are going to struggle to keep even 60fps at 4k.
 
The temps on the card have nothing to do with the actual heat being put out. Using your logic then a passive cooled gtx 1050 at 75 C would "heat" up your room more than a gtx 1080 Ti at 65 C...

Well I'm sure he meant TDP in which case a 350W monster pushing 80C will heat up a room very fast.
 
Cherry picked site I take it. No way in hell I would believe those rankings as fact.
Love your logic mate. "results don't support my my perception of performance landscape, must be biased site".
 
Green has been killing in performance per Watt since Maxwell compared to Red. I love my AMD but I build almost all SFF boxes now and the power/heat situation just favors nvidia. I still am willing to pick up a 1070 tier vega card if I can run it with my sf450 corsair psu and Ryzen 1700. If not I KNOW a gtx 1080 will drop right in. Not long to see how it all shakes out.
I'm running a 1080Ti + 6700k on a Silverstone ST45SF (older and less efficient SFX SPU) without issue. The SF450 can easily handle any single GPU currently on the market unless you start pushing OCs.
 
myth. maybe 970 vs 390. but 980ti was always around fury x power consumption and much higher with OC. a 980 would go over 200W with boost on the custom ones and a 290x could be between 200-250W. Not as simply in favor of nvidia as people claim. The main change came with pascal

Agreed. High end maxwell was thirsty. I started looking at Nvidia again when the 970 hit and ultimately picked up a 960 cheap. Obviously a different tier etc. Really considered the 290/390Fury but didn't want to shell out. 970 power draw for my then performance target was attractive but again went down to the 960. Pascal Is definitely more efficient for sure. Had a 1070 sc. Awesome card. Sold it for a small profit. I really want a Vega gpu for my Ryzen box but depending on price/ availability may cave and get a used 1080.
 
Hmm. I base my gpu upgrade cycle on graphs like the ones above. Once my current gpu is no longer even included, it's time to upgrade. Sigh.
 
I am bent on a vega card for the right price. dont want to give up freesync and i am curious where i can go with undervolting/downclocking. if the price works out with all that I am game. not doing 4k so destroying 1440p will be enough. for the right price.

So with Freesync you ended up locking yourself in with an inferior outdated inefficient card. Sounds great :)

Performance a year ago now at 300W!

When GV106 launches it seems Vega cards will be sub 250$.
 
Mainstream performance at very high powerdraw. Sounds like very limiting to me.
 
Back
Top