• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

News AMD Announces Radeon VII

Page 24 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Veradun

Senior member
Jul 29, 2016
306
244
86
Without taking your (or anyone else's) words or posts out of context, Vega 56 is exactly where it was at launch as it is today in relation to it's performance vs. it's main competition.

VS. the 1070 at launch (6.4% faster than a 1070)


VS. the 1070 today (7.5% faster than a 1070)
They don't retest older cards, it can't stand anywhere else then where it was in the beginning :>
 

Heclone

Junior Member
Dec 7, 2018
19
11
41
I would add that the TPU test protocol mostly use DX11, and that's partially why they get that bad results for Vega regarding others websites.
 
Mar 11, 2004
19,214
1,750
126

Radeon VII running at 999mV, the efficiency is pretty good at ~1800MHz compared to Vega 64 LC, it just needs a better cooler from a board partner.
That seems decently impressive and would be more what I'd expect was going to be possible with Vega 20 (although I still feel like it isn't terribly impressive, but that certainly would make it not almost plain disappointing like it is at stock due to the power, thermals, and noise for the performance it offers). People in the CPU forum were trying to gauge 7nm based on stock Radeon VII which is silly because of how out of whack AMD's GPU voltages are (where its not unheard of to gain performance/clocks by undervolting some). But if 1800MHz at ~1V is fairly easy/common for Vega 20, that looks like it'd be cutting power by possibly close to 1/3, with no performance loss (and much better heat/temp and thus noise).

What is providing the readouts? I'd like something more in depth and comprehensive (even just a kill-a-watt, but ideally something better for power measurement, and then do a sustained like 2-3 hour run of gaming and compare the power, temp, clocks, and performance with stock - and if possible multiple voltage levels so we can see if there's a sweetspot that might be more feasible for most users). And one for compute loads (since I think Vega is better utilized during those tasks).

Would also be interesting to see where it could get with a good custom water loop, and see if it could be pushed close to 2GHz and how the power and performance would be there. Seems like the card has the power setup to see just what it can achieve.
 

Adawy

Member
Sep 9, 2017
79
23
41
That seems decently impressive and would be more what I'd expect was going to be possible with Vega 20 (although I still feel like it isn't terribly impressive, but that certainly would make it not almost plain disappointing like it is at stock due to the power, thermals, and noise for the performance it offers). People in the CPU forum were trying to gauge 7nm based on stock Radeon VII which is silly because of how out of whack AMD's GPU voltages are (where its not unheard of to gain performance/clocks by undervolting some). But if 1800MHz at ~1V is fairly easy/common for Vega 20, that looks like it'd be cutting power by possibly close to 1/3, with no performance loss (and much better heat/temp and thus noise).

What is providing the readouts? I'd like something more in depth and comprehensive (even just a kill-a-watt, but ideally something better for power measurement, and then do a sustained like 2-3 hour run of gaming and compare the power, temp, clocks, and performance with stock - and if possible multiple voltage levels so we can see if there's a sweetspot that might be more feasible for most users). And one for compute loads (since I think Vega is better utilized during those tasks).

Would also be interesting to see where it could get with a good custom water loop, and see if it could be pushed close to 2GHz and how the power and performance would be there. Seems like the card has the power setup to see just what it can achieve.
Yeah, that's why I said "compared to Vega 64 LC". The card being pushed to 2GHz and beyond, is what I want to see next.
 

SpaceBeer

Senior member
Apr 2, 2016
291
95
71
So the pro card buyers look silly now?
Unbelievable. There is really no need to buy any WX card above WX5100 now. I was thinking of buying WX 4100 /5100. But if Vega VII will continue to have Pro drivers and certification for major CAD programs, it is without doubts the best possible purchase
 

Ajay

Diamond Member
Jan 8, 2001
5,670
1,652
136
This doesn’t make any sense. Why would AMD throw away the extra revenue they’d earn with the MI50?!
 

AtenRa

Lifer
Feb 2, 2009
13,064
1,882
126
MI50/60 are datacenter/server cards, they have features missing from desktop Radeon VII , such as IF Links and MxGPU for up to 8 GPUs per VM. Also MI60 has 32GBs of memory with 64 CUs and both MI50 and MI60 has higher FP64 performance.

They dont take away revenue from the MI50 cards (datacenter) by selling the Radeon VII as Pro card.
 
  • Like
Reactions: lightmanek

SpaceBeer

Senior member
Apr 2, 2016
291
95
71
They take away revenue from WX cards. Why would anyone buy WX 8200/WX9100 when Vega VII is both cheaper and better. Even WX 7100 at ~500€ doesn't seem so attractive. I mean, if I'm spending 500€ for Pro card, I would rather invest 50% more money to get 100% better product.

EDIT:
Ok, it's not what I thought. Apparently, all AMD cards will have support fro Pro drivers, but only Pro cards will be optimized for professional workload and will have application certifications

https://community.amd.com/community/radeon-pro-graphics/blog/2019/02/11/amd-radeon-pro-software-for-enterprise-19q1-delivers-solidworks-performance-improvements-and-over-320-day-zero-certifications
 
Last edited:

Hitman928

Golden Member
Apr 15, 2012
1,931
744
136
They take away revenue from WX cards. Why would anyone buy WX 8200/WX9100 when Vega VII is both cheaper and better. Even WX 7100 at ~500€ doesn't seem so attractive. I mean, if I'm spending 500€ for Pro card, I would rather invest 50% more money to get 100% better product.

EDIT:
Ok, it's not what I thought. Apparently, all AMD cards will have support fro Pro drivers, but only Pro cards will be optimized for professional workload and will have application certifications

https://community.amd.com/community/radeon-pro-graphics/blog/2019/02/11/amd-radeon-pro-software-for-enterprise-19q1-delivers-solidworks-performance-improvements-and-over-320-day-zero-certifications
Relevant Graphic from the link:
 

zinfamous

No Lifer
Jul 12, 2006
97,484
12,038
136
Now I wonder if they are going to ship more than 5k or 10k units, whatever it is. Pro users are still going to require support and certifications for the Pro hardware, but this definitely makes the Radeon VII way more attractive at that price. Seems kinda pointless to unlock this feature with such a limited number of boards, no?
 

Hitman928

Golden Member
Apr 15, 2012
1,931
744
136
Now I wonder if they are going to ship more than 5k or 10k units, whatever it is. Pro users are still going to require support and certifications for the Pro hardware, but this definitely makes the Radeon VII way more attractive at that price. Seems kinda pointless to unlock this feature with such a limited number of boards, no?
If you read the fine print, it looks like the limited Pro drivers will be available on any GCN graphics card.

 

Mopetar

Diamond Member
Jan 31, 2011
4,539
743
126
RTX cards don't even do Ray Tracing that well...why address it?
That's the headline that the article ran with. I don't think this has much of anything to do with ray tracing. It is free media coverage though.

Unbelievable. There is really no need to buy any WX card above WX5100 now.
Availability, since if you can't buy a Radeon VII if they're all sold out and AMD doesn't make any more of them.

This doesn’t make any sense. Why would AMD throw away the extra revenue they’d earn with the MI50?!
They needed a 7nm GPU for CES. The most recent rumors are suggesting that Navi won't launch until October/November instead of June/July, so it may have more issues than initially thought. Some rumors suggested that they were only going to ship 20k Radeon VII GPUs at most, so it's just a matter of how much they value any publicity they get from releasing it as opposed to the value they could have from selling the cards as MI50 products.
 

ASK THE COMMUNITY

TRENDING THREADS