llien
Member
Didn't GN also confirm his "70-100MH/s"?
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
Radeon RX Vega thread
- Thread starter PFD
- Start date
Didn't GN also confirm his "70-100MH/s"?
Indeed they did but that still leaves another etailer and the fact an AMD rep couldn't give a straight answer.
Unknown Soldier
Member
AMD with their own twist on the Founders Edition. At least this means people can stop complaining about Founders Edition now since both competitors are doing it.
Reallink
Member
Back in May/June, Newegg ran multiple sales moving those 1080's on their Ebay store front for $389 - $429. They were brand new shipped by Newegg direct.
ethomaz
Banned
It should complain on both cases.
IMACOMPUTA
Member
Wasn't Nvidia transparent at least? We knew that the Founder's Edition cards were marked up.
Also, this seems like an actual marketing trick to appear to have more value.
Nvidia just charged an impatience tax.
Unknown Soldier
Member
This is correct. Nvidia said straight up, pay more now for reference cards or wait a few weeks for the AIBs. They were completely up front about it. There was never any subterfuge or secret price raises. What AMD is doing is actually considerably more underhanded to the consumer.
michaelius
Banned
Not really - there's a huge difference between deception (remember reviewers were also told it's 500$ and made reviews in that context) and early adopter tax.
ethomaz
Banned
thelastword
Banned
RX Vega 56 vs GTX 1070 @ 1080p
https://www.youtube.com/watch?v=8vTI_xrEduc
https://www.youtube.com/watch?v=t6AcLbe5uTw
https://www.youtube.com/watch?v=8vTI_xrEduc
https://www.youtube.com/watch?v=t6AcLbe5uTw
Not exactly, as the aib cards outside of evga were pretty much all out the founders price or higher, for several months. Evga lower priced cards were sold out for months too. Most 1070s at retail are also still at founders level price if you can find one. Imo it was still largely deceptive of Nvidia.
Beautiful Ninja
Member
Nvidia didn't set those prices though, AIB partners and retailers did, as they saw Pascal cards were super popular and could sell at a higher than MSRP price.
AMD is setting these new prices after an intentional attempt to deceive reviewers with lower Vega pricing to take account for in their reviews. AIB partners and retailers have to follow suite or lose money.
dr_rus
Member
And all 6 or those are AMD sponsored as well as Total War which they've spent half a year trying to get faster than NV's DX11 in DX12 but just couldn't make it. So 7 to 2 there are AMD sponsored games. And this is what you're likely using for your conclusions on how AMD cards are aging better these days. It's just pure BS.
Looks like it won't. Next GF will be based on a newer architecture than Volta.
1080Ti FE cost the same as 1080Ti AIB.
They were also all OCed and using better coolers than FEs which means that they would've cost the same even if the FE price would be lower.
Whoa, that's news to me.
Edit: I googled and read what the CEO recently said at the earnings call.
FingerBang
Member
Pretty much. And this is basically the only vega that makes sense to buy (especially if paired with a freesync display).
Coulomb_Barrier
Member
Indeed it seems Nvidia's next line of GPUs won't be arriving until Feb-May 2018 after CEO's comments.
So those wishing the Vega's to get blown out the water by Pascal's successor will be waiting a long time. Looking like at least 6 months! Plenty of time for both Vega's to clearly overtake their direct Pascal equivalents in perf.
Problem with Volta it seems is its very expensive to manufacture. Expect prices to rise again (+$50 across the line-up).
So those wishing the Vega's to get blown out the water by Pascal's successor will be waiting a long time. Looking like at least 6 months! Plenty of time for both Vega's to clearly overtake their direct Pascal equivalents in perf.
Problem with Volta it seems is its very expensive to manufacture. Expect prices to rise again (+$50 across the line-up).
Yeah, that would really be an issue if Pascal wasn't already blowing Vega out of the water.
Also, it's a bit amusing to talk about waiting "a long time" of "at least 6 months" after we just had a release which ended a period of 15 months of waiting for something from AMD to kind-of sort-of catch up to NV's May 2016 releases.
(Yeah, if you can't tell I'm a bit annoyed at just how unimpressive Vega is, especially since I have to buy one)
Have to? For dev purposes? Cause didn't you get a 1080 last year?
Yes, for testing. (And yes, I got a 1080 for my VR setup, which was a really good decision)
Delusibeta
Banned
Won't happen with the Vega 64, it's constantly behind the 1080, and the gap is far too large for AMD's moneyhatting of new games to change the picture in a mere six months.
thelastword
Banned
Vega 64 vs GTX 1080 @ 1440p
https://www.youtube.com/watch?v=J2fMCgyQhkg
Vega 56 vs GTX 1070 @ 1440p
https://www.youtube.com/watch?v=UgIFWcRLKio
https://www.youtube.com/watch?v=uzbjc2JmQJM
Vega 56 vs GTX1070 tested in 25 games 1080p/1440p/2160p
https://www.youtube.com/watch?v=cjH08ncGxEY
https://www.youtube.com/watch?v=J2fMCgyQhkg
Vega 56 vs GTX 1070 @ 1440p
https://www.youtube.com/watch?v=UgIFWcRLKio
https://www.youtube.com/watch?v=uzbjc2JmQJM
Vega 56 vs GTX1070 tested in 25 games 1080p/1440p/2160p
https://www.youtube.com/watch?v=cjH08ncGxEY
michaelius
Banned
Will be funny if it turns out V56 is price competitor to 1080.
Coulomb_Barrier
Member
You lucky boy I would personally be pleased to buy a Vega 56 once the prices settle. £380 for a custom this time next month and I'll be there. If not, life goes on and I wait until July 2018 for a GeForce 2070.
Dude, when every word in multiple paragraphs is underlined it becomes an unreadable mess.
thelastword
Banned
Vega 56 gets pretty close to a GTX1080 and beats it in several games, especially more recent games, so I can see it selling for up-to $500-550.00 from AIB's.
OTOH
Apparently, AIB water cooled blocks will do a lot for Vega as far as cooling solutions go. The current beta driver seems to be downclocking HBM on the Vega FE, don't know if that translates to the RX, but there's a lot of driver optimizations and a lot of perf not utilized on Vega right now...
HBM does not seem to use more than 20watts, that's why adding +50 to your power limit is free performance on Vega cards, as HBM still does not go beyond 20 watts in that case. If that limit could be offset and also the voltage limits increased by AIB's then we will see some interesting results. So far, Vega 56 can be pushed to 1.3V so you will get overclocks to HBM in the 950-980Mhz range. The Vega FE goes up to 1.35 and you get overclocks up to 1100Mhz. All preliminary stuff of course, but a bios unlock mode will open up lots for Vega, so I hope AMD obliges...
There has not been much tinkering with Vega 64, but I would assume it's voltage limits should be higher than 56, in that case, higher overclocks on HBM along with core clocks will bring alot of additional perf to that card with a good cooling solution.
llien
Member
Yep.
I wonder if AMD can really produce enough Vega 56s to satisfy market demand at $399.
Source?
thelastword
Banned
The post is just above, it reads...
"Vega 56 gets pretty close to a GTX1080 and beats it in several games, especially more recent games"
Dirt 4, Warhammer, The Division, Infinite Warfare, Civilization and Ashes it matches or beats the 1080 with a slight OC. I'm just saying with a better cooler and AIB cards you can expect even better performance....where you will see the 56 beating GTX 1080 in more games on top of driver optimizations....
For reference, here's the delta between a stock vega 56 (without an OC) against a gtx 1080. Gtx 1080 has a bigger lead in more older games than newer ones. I don't think 56 will ever beat or match 1080 in GTA etc....but it has a chance to match and beat it in many others with AIB cards...
Just check the delta on some of the games.....Obviously 1080 has a big lead in some, but in others not so much...
Vega 56 vs GTX 1080 @ 1080p
https://www.youtube.com/watch?v=A9IbkDoRwvg&t=2s
In essence, I can see the Vega 56 surpassing or matching the GTX 1080 in these games with a decent AIB card. More surpassing than matching tbh as it's already matching the 1080 in a few games already.
Far Cry Primal
Resident Evil 7
Hitman
Deus Ex
Titanfall 2
Total War Warhammer
Civilization
Watch Dogs 2
Infinite Warfare
Quake Champions
Sniper Elite 4
Dawn of War
The Dirt lead will extend
Match it in DOOM with 1% lows
Match it or come close in Witcher 3, Battlefied 1, Mirrors Edge, Dihonored 2 and For Honor....
The bigger lead I see for the GTX1080 that will not be matched is GTAV, Crysis 3, Overwatch etc...I think it's too late for AMD to tackle gains there, but who knows with Crysis 3 as the results are a bit odd. I also happen to believe that Crysis performs better with an 8 core ryzen over a 7700k, so we'll see...PUBG is in beta, but it's lead on NV is pretty big, so we'll see how that pans out.
OTOH
Why aren't we seeing some MS games like Gears of War 4, Gears of War Remaster, Recore, Quantum Break, Forza 6, Forza Horizon 3, Killer Instinct in some of these benchmarks.. Vega 56 vs 1070? Framerates from 1080p-2160p would be interesting information.......It will also be interesting to see scores on games like Batman Arkham Knight, Bioshock Infinite/Remasters, Skyrim remaster etc....Tekken 6 is yet another...
llien
Member
ASUS Strix Vega 64 preview (CB)
https://translate.google.com/transl...x-vega-64-vorserie-test/2/&edit-text=&act=url
So, a bit better perf/watt, closer to "turbo" mode performance, about 310W consumption when gaming vs 210W on AIB 1080.
Very quiet.
https://translate.google.com/transl...x-vega-64-vorserie-test/2/&edit-text=&act=url
So, a bit better perf/watt, closer to "turbo" mode performance, about 310W consumption when gaming vs 210W on AIB 1080.
Very quiet.
Lagspike_exe
Member
This looks very good, aside from TDP.
ethomaz
Banned
Vega 64 didn't beat GTX 1080 so how Vega 56 do that?
Newer games (in your chosen resolution):
Resident Evil 7 = +0%
Ghost Recon Wildlands = +23%
Snipe Elite 4 = +15%
Prey = +16%
Styx: Shards of Darkness = +24%
GTX 1080 has a big lead in newer games... old games like Doom has big advantage for Vega.
Using the games you listed (in your chosen resolution):
Warhammer 40,000: Dawn of War III = +16%
Civilization VI = +29%
Call of Duty: Infinite Warfare = +0%
Ashes of the Singularity = +17%
Hitman = +14%
Deus Ex: Mankind Divided = -3%
Watch Dogs 2 = +29%
Total War: Warhammer = +40%
PS. I'm a bit lazy to look for a 3rd review site to show The Division and Dirt 4 results
Edit - You did list more games... most of them with GTX 1080 lead... you are trying to make Vega 56 has a performance of something it doesn't have... perhaps you are confusing Vega 56 as Vega 64.
Hexus has a "review" on the Asus Strix Vega 64 as well:
http://hexus.net/tech/reviews/graphics/109078-asus-radeon-rx-vega-64-strix-gaming/
Looks pretty good, power draw++ though
http://hexus.net/tech/reviews/graphics/109078-asus-radeon-rx-vega-64-strix-gaming/
Looks pretty good, power draw++ though
dr_rus
Member
GV100 is 150% of GP100 in size so of course it's more expensive to manufacture. But this has no relation to Volta (architecture) or anything coming for GeForce markets.
Also, yeah, I'm pretty sure that the gap between Vega's launch and the next GeForce lineup will be smaller than that between Pascal and Vega. So you could say that Vega has about those 6 months or so to "to clearly overtake" Pascal in those three to five AMD sponsored games with botched low level renderers.
Locuza
Member
Now Gamestar testet the HBCC too.
The results are the worst from all three (PCGH, Computerbase, Gamestar).
The HBCC is always slower or just as fast as off.
On average HBCC off was 4-5 faster:
http://www.gamestar.de/artikel/radeon-rx-vega-64-benchmarks-mit-high-bandwith-cache-controller,3318564.html
PCGH found that the HBCC was always as fast or faster than without.
CB had mixed results, sometimes the HBCC won, sometimes it lessened the performance.
The games each one testet were different and so was the testsystem.
PCGH used the 6800K @ 4,4 Ghz with 32GB (3200) in quad-channel-mode, ~4GB system memory were reserved for ~12 GB unified memory.
CB used also 32GB (probably quad-channel (3000) with the 6850K @ 4,3 Ghz)
They reserved 8GB for 16GB unified memory.
Gamestar in contrast used the 7700K @ default with 16GB (2400) dual-channel memory where they reserved 4GB for 12 GB unified memory.
The results are the worst from all three (PCGH, Computerbase, Gamestar).
The HBCC is always slower or just as fast as off.
On average HBCC off was 4-5 faster:
http://www.gamestar.de/artikel/radeon-rx-vega-64-benchmarks-mit-high-bandwith-cache-controller,3318564.html
PCGH found that the HBCC was always as fast or faster than without.
CB had mixed results, sometimes the HBCC won, sometimes it lessened the performance.
The games each one testet were different and so was the testsystem.
PCGH used the 6800K @ 4,4 Ghz with 32GB (3200) in quad-channel-mode, ~4GB system memory were reserved for ~12 GB unified memory.
CB used also 32GB (probably quad-channel (3000) with the 6850K @ 4,3 Ghz)
They reserved 8GB for 16GB unified memory.
Gamestar in contrast used the 7700K @ default with 16GB (2400) dual-channel memory where they reserved 4GB for 12 GB unified memory.
ethomaz
Banned
Maybe HBCC depends of the system RAM performance but even the best review until now showed little to no gain with HBCC.
kirankara78
Member
Wouldn't buy a blower style card for any amount of money. Wait and pay extra for better cooler
Locuza
Member
PCGH showed a little teaser for next week (unknow topic so far).
But they testet a game where Nvidia ist CPU limited and a 1060-1080Ti are pretty much at the same level.
Interestly enough, Vega 56 and Vega 64 are beating all Geforces.
So in this CPU limited case AMD is in front, which is quite rare.
Hawaii (390X) might perform worse than expected, Fury X is just a bit faster than the 580 but Vega really pulls off there.
But increasing the resolution to UHD absolutely tanks the Radeon performance, Phil from PCGH (blaidd in 3DCenter) says that the Vega 56 is roughly performing on 1060 GTX level, Vega 64 ~ 1070 GTX.
The game could be PLAYERUNKNOWN'S BATTLEGROUNDS.
https://www.forum-3dcenter.org/vbulletin/showthread.php?p=11465009#post11465009
But they testet a game where Nvidia ist CPU limited and a 1060-1080Ti are pretty much at the same level.
Interestly enough, Vega 56 and Vega 64 are beating all Geforces.
So in this CPU limited case AMD is in front, which is quite rare.
Hawaii (390X) might perform worse than expected, Fury X is just a bit faster than the 580 but Vega really pulls off there.
But increasing the resolution to UHD absolutely tanks the Radeon performance, Phil from PCGH (blaidd in 3DCenter) says that the Vega 56 is roughly performing on 1060 GTX level, Vega 64 ~ 1070 GTX.
The game could be PLAYERUNKNOWN'S BATTLEGROUNDS.
https://www.forum-3dcenter.org/vbulletin/showthread.php?p=11465009#post11465009
I would mount another cooler, for reference i was fine with the RX 480 blower cooler in my mATX case.
kirankara78
Member
Fair enough. Tbh I used to have a 690 years ago which never gave me issues, but it was little noisy and ran hot.
In theory blowers can be useful, so I should have asked him what set up he had and a blower might be useful for him in reality.
But overall blowers aren't really best option in most scenarios
kirankara78
Member
http://www.gamersnexus.net/industry/3025-amd-statement-on-vega-pricing-retailers-fault
Seems there might be truth to amd being less than honest about this.
Sorry if this already been posted
Seems there might be truth to amd being less than honest about this.
Sorry if this already been posted
Nydus
Gold Member
No its not. Your Screenshot shows 1080p and those framerates are way too low. My strix 1080ti has those framerates @1440p with everything @ ultra in yasnaya (or whatever it is called). Its already confirmed that AMD is atm not so good in pubg.
Edit: a sorry missed the part with cpu limit. Cpu limit with a 6900k??? I make a quick test with the ti @1080p and report back.
And on the pricing lie:
I think thats really scummy. I hope the techpress updates their reviews if true
ISee
Member
BTW; We can also expect prices for Polaris and Pascal to further increase. Samsung wants to concentrate a bit more on NAND flash and wants to cut down GDDR production. They expect GDDR5 prices to go up by ~30%. Now I don't know how expansive they are in the first place, but we can expect at least a slight price bump for GPUs, again.
I think i might just order the stock card if i can grab it cheap
The store here got 45 days return policy and i wont be able to use the card in that time period anyway, if the price on the non-ref cards are fine i will just return the other card.kirankara78
Member
I've heard it amounts to couple dollars per chip
" Digitimes says memory prices for graphics cards jumped more than 30 percent in August. That translates into a $2 jump in the average price, which went from $6.50 in July to $8.50 in August, so we are not talking about astronomical increases here. Still, it adds up. For example, a GeForce GTX 1050 is equipped with four memory chips, a GeForce GTX 1060 has six memory chips, and higher end cards like the GeForce GTX 1070 use eight chips."
Ouch