Gráfica AMD Vega (Radeon Vega 56 & 64 & Frontier Edition)

[MDK]

Não te esqueças que a VEGA ou a BIG VEGA vem com memória HBM2 e não GDDR5 como as 1080TI

Lol e o quê que isso têm haver ! ?? o que me parece é que como na geração da fury X o HBM foi um se não o maior buraco da fury X
para uma tech que ta morta já vai sair o GDDR6 o HBM para large scale ainda não dá só a AMD não percebe isso, a mim nem a ninguém interessa os numeros que me interessa se tens 1 milhão de bits faz menos FPS menos performance e experiência inferior , se fato de trazer HBM fosse grande vantagem tinha mostrado isso só mostrou foi o contrario como já referi, Se for melhor e mais performance tiver já disse sou 1º como habitual a comprar e as primeiras fotos aparecer no forum ZWAME como habitual também das primeiras Vega day one a serem vendidas, mas a jogar pelo habitual fico aguardar não tenho muita Fé

cheers

 

[ObscureAngel]

@MDK O problema acho que não estava no HBM, mas também concordo que isso foi uma grande desvantagem para a AMD.
Embora demonstre que a 4K nao perde tantos frames comparado com o GDDR5, os frames finais que debita é que conta, also, a pouca quantidade de vram nao ajudava.
Para além que a fraca disponibilidade do HBM e o preço do mesmo só piorava ainda mais.

Mas a meu ver os maus resultados prendiam-se mais por a AMD ter quase 0 a nível de compressão de memória, ya atiravam 500GB/s ou algo do genero, mas ia a nvidia com 320gb/s ou menos e conseguia os mesmos resultados em memory benchmarks.
Agora a AMD já investiu um pouco mais em memory compression, ainda não está nem lá perto ao nível da nvidia, mas com aquela cache por hardware, ainda estou para ver os resultados daquilo.

Porque no fundo vejo a AMD a atirar-se com muitas specs, muita força bruta, e embora a otimização tenha estado a aparecer lentamente com as ultimas revisões GCN, a sensação que me dá sempre, é que eles precisam sempre de muita lenha para andar, e a nvidia com muito menos specs e mais eficiência, fazem o mesmo.

 

[Melli]

Não te esqueças que a VEGA ou a BIG VEGA vem com memória HBM2 e não GDDR5 como as 1080TI

E isso é relevante porque...? Mesmo ignorando o facto que a 1080ti não usa gddr5, o que mais conta é a banda total.

 

[THESLB]

https://www.overclock3d.net/news/gp..._specifications_leak_on_compubench_database/1 Agora já são 16gb vram

 

[black_edition]

Entire AMD Vega Lineup Reportedly Leaked - Available on June 5th?

Reports are doing the rounds regarding alleged AMD insiders having "blown the whistle", so to speak, on the company's upcoming Vega graphics cards. This leak also points towards retail availability of Vega cards on the 5th of June, which lines up nicely with AMD's May 31st Computex press conference. An announcement there, followed by market availability on the beginning of next week does sound like something that would happen in a new product launch.

On to the meat and bones of this story, three different SKUs have been leaked, of which no details are currently known, apart from their naming and pricing. AMD's Vega line-up starts off with the RX Vega Core graphics card, which is reportedly going to retail for $399. This graphics card is going to sell at a higher price than NVIDIA's GTX 1070, which should mean higher performance. Higher pricing with competitive performance really wouldn't stir any pot of excitement, so, higher performance is the most logical guess. The $399 pricing sits nicely in regards to AMD's RX 580, though it does mean there is space for another SKU to be thrown into the mix at a later date, perhaps at $329, though I'm just speculating on AMD's apparent pricing gap at this point.

Next up is the RX Vega Eclipse, which will reportedly retail for $499, going head to head with NVIDIA's GTX 1080 (in fact, slightly cheaper than the majority of AIB versions of the card). The line between the Core and the Eclipse is a little blurry here, since we know that the GTX 1070's performance can easily be overclocked to reach a stock 1080 - their performance delta isn't that great. If the RX Vega Core does bring with it higher performance than the GTX 1070 (justified by its higher pricing), then that would place it close to GTX 1080 (stock) performance. Since AMD would be trying to avoid its RX Vega Core from eclipsing (eh) its RX Vega Eclipse graphics card in the price/performance department, one can expect - with reservations - that the performance delta between the Core and the Eclipse is higher than their respective pricing indicates. So I would expect the RX Vega Eclipse to offer performance that's greater than the GTX 1080's.

Finally, we have the crème de la crème of Vega, the RX Vega Nova. This graphics card is reported to retail for $599, a full $100 cheaper than NVIDIA's GTX 1080 Ti, while looking to directly compete with it. Considering this pricing, and admitting that the leak pans out correctly, this would mean we won't be seeing a Vega card that's capable of competing with NVIDIA's Titan Xp graphics card (at least, not a single-GPU solution...) AMD simply would not sell their top of the line Vega for $599 if it was competitive with that NVIDIA titan of a graphics card. Based on AMD's previous pricing strategy, I would expect the company to deliver roughly the same performance as the GTX 1080 Ti, looking to use its Nova not as a pure performance product, but as a price/performance contender. What do you make of this leak?

 

[Miguel_Pereira]

1600mhz parece-me bem decente, depois dos que vimos com as RX4/580. Caso seja mesmo verdade, pode ser que as vega venham a ser mesmo um bom lançamento!

https://videocardz.com/69475/amd-radeon-vega-spotted-with-16gb-memory-and-1600-mhz-clock

 

[ruben-moniz]

Fazendo contas usando matematica "hype-train" dá Vega=(1.19*1.78*1.05)*RX580 = 1.05*1080ti | [1.19 frequencia, 1.78 CU, 1.05 melhoramentos de arquitetura].

 

[Miguel_Pereira]

Faz sentido, espera-se que a vega possa concorrer com o GP102. Mas ainda é mt hype-train. Mais vale esperar pelo lançamento.

 

[muddymind]

Parece que hoje vamos ter mais um teaser sobre as vega: https://www.techpowerup.com/233352/amd-executives-tease-vega-reveal-on-todays-event

Já começa a fartar tanto teaser e promessa e depois não passar daí...

 

[MikeAVC]

Ultrapassar as placas da Nvidia por um preço mais baixo é a única maneira que vejo de compensar o tempo de espera. Espero que a nova arquitectura resolva as limitações do GCN 1-4 como a clockspeed, e seja o Maxwell da AMD.

 

[vandaminator]

Os clocks pouco interessam, desde que tenha boa performance. Até porque já não é GCN

 

[ObscureAngel]

@vandaminator Ainda é GCN.
Apenas é a primeira vez em 4 revisões que vão finalmente optimizar os GCN cores, coisa que a nvidia faz em práticamente todas as séries novas com os CUDA Cores.

 

[VelhoDoRestelo]

Ultrapassar as placas da Nvidia por um preço mais baixo é a única maneira que vejo de compensar o tempo de espera. Espero que a nova arquitectura resolva as limitações do GCN 1-4 como a clockspeed, e seja o Maxwell da AMD.

Duvido que tal aconteca... Certamente sera para combater as mesmas, mas duvido que sejam realmente superiores e a precos mais baixos!!!

 

[muddymind]

Temos então a versão profissional:

Guru3D:

AMD launches Radeon Vega Frontier Edition

AMD just launched their Radeon Vega Frontier Edition card. It is a 'mission intelligent' enterprise graphics card for professional usage (data-crunching) and product designers, and yes it is not for PC gamers.

The card comes with 16 GB of HBM2 graphics memory and will perform in the 13 TFLOP (fp32) performance bracket. It will be available late June. At this point there have been no consumer announcements regarding Radeon RX Vega graphics cards. From the looks of it, the announcements will be made during Computex.

TechPowerUp:

AMD Announces Radeon Vega Frontier Edition - Not for Gamers

Where is Vega? When is it launching? On AMD's Financial Analyst Day 2017, Raja Koduri spoke about the speculation in the past few weeks, and brought us an answer: Radeon Vega Frontier Edition is the first iteration of Vega, aimed at data scientists, immersion engineers and product designers. It will be released in late June for AMD's "pioneers".

This news comes as a disappointment to all gamers who have been hoping for Vega for gaming, because it reminds of what happened with dual Fiji. A promising design which ended up unsuitable for gaming and was thus marketed for content creators as Radeon Pro Duo, with little success. But there is still hope: it just looks like we really will have to wait for Computex 2017 to see some measure of details on Vega's gaming prowess.

Vega Frontier Edition is the Vega GPU we've been seeing in leaks in the last few weeks, packing 16 GB of HBM2 memory, which, as we posited, didn't really make much sense on typical gaming workloads. But we have to say that if AMD's Vega truly does deliver only a 1.5x improvement in FP32 performance (the one that's most critical for gaming at the moment), this probably paints AMD's Vega as fighting an uphill battle against NVIDIA's Pascal architecture (probably ending up somewhere between GTX 1070 and GTX 1080). If these are correct, this could mean a dual GPU Vega is indeed in the works, so as to allow AMD to reclaim the performance crown from NVIDIA, albeit with a dual-GPU configuration against NVIDIA's current single-chip performance king, Titan Xp. Also worth nothing is that the AMD Radoen Vega Frontier Edition uses two PCI-Express 8-pin power connectors, which suggests a power draw north of 300 Watts.

For now, it seems AMD actually did its best to go all out on the machine learning craze, looking for the higher profits that are available in the professional market segment than on the consumer side of graphics. Let's just hope they didn't do so at the expense of gaming performance leaps.

After an initial throwback to AMD's times since he became lead of Radeon Technologies Group, where Raja mentioned the growing amount of graphics engineers in AMD, including their commitment to the basics of graphics computing: power, performance, and software. Better basics in hardware, software, and marketing are things that Raja says are responsible for AMD's current market outlook, both from a gamer and content creator perspective, which led to an increase in AMD's graphics marketshare.

RTG's chapter two of Radeon Rising, going beyond the basics, will allow the company to go after premium market dollars, with an architecture that excels on both gaming and CAD applications. Raja Koduri said he agreed with NVIDIA CEO Jensen Huang in that at some point in the future, every single human being will be a gamer.

The final configuration of Vega was finalized some two years ago, and AMD's vision for it was to have a GPU that could plow through 4K resolutions at over 60 frames per second. And Vega has achieved it. Sniper Elite 4 at over 60 FPS on 4K. Afterwards, Raja talked about AMD's High Bandwidth Cache Controller, running Rise of the Tomb Raider, giving the system only 2 GB of system memory, with the HBCC-enabled system delivering more than 3x the minimum frame-rates than the non-HBCC enabled system, something we've seen in the past, though on Deus Ex: mankind Divided. So now we know that wasn't just a single-shot trick.

Raja Koduri then showed AMD's SSG implementation and how it works on a fully ray-traced environment, with the SSG system delivering much smoother transitions in the system. AMD worked with Adobe on integrating SSG capability into Adobe Premiere Pro.

Raja then jumped towards machine intelligence, which Raja believes will be dominated not by the GPU (NVIDIA's green) or CPU (Intel blue) paths, but in true heterogeneous computing.

Raja took to stage results on DeepBench, a machine learning benchmark where NVIDIA dominates at the moment, joking about AMD's absence from the benchmark - since they really didn't have a presence in this area. In a benchmark, AMD pitted Vega against NVIDIA's P100 architecture (interestingly, not against NVIDIA's recently announced V100 architecture, which brings many specific improvements to this kind of workloads), delivering an almost 30% performance lead.

 

[THESLB]

Entao mas se essa versao sai em junho para PRO´s,as versoes para gamers só saiem quando?no natal?

 

[wepsilon]

Pelos vistos vamos ter de esperar 2 semanas para o anúncio oficial e concreto dos VEGA, no Computex Taipei 2017 com início a 30 de Maio.
I hope!

 

[Dark Kaeser]

Não houve grande novidade, já se sabia do novo controlador de memória HBCC, que já se sopunha ter algo a haver com o, também já anteriormente apresentado, SSG (SSD on Graphics), novidade é mais a integração do software, que para mercados profissionais sempre foi calcanhar de aquiles da AMD, apesar de estar a recuperar terreno nos últimos anos. Sinal disso é o trabalhar com a Adobe para integrar essa capacidade do SSG no Premier Pro, outros virão certamente.

 

[Olissipo]

Houve uma sessão de perguntas com o Raja Koduri acerca das Vega no /r/AMD.

Um user fez um resumo:

• Main goal of Vega was to create an architecture that can handle large data sets and game at 4K at 60fps.
• The demo during Financial Analyst day was an air-cooled Frontier Edition, not an RX Vega card. There will be water cooled versions of Vega that will run slightly faster.
• Frontier uses 2 x 8GB stacks of HBM2.
  
• Both HBM1 and HBM2 provide plenty of bandwidth.
  
• Raja will look into OC'ing HBM2 and a 16GB RX card.
• RX Vega will be shown at Computex. It will not be available the same week but nothing else has been ruled out.
• Frontier runs comfortably using 1x6-pin and 1x8-pin but RTG put 2x8-pin on the production card for more headroom.
• Raja is keeping his beard until Vega launches.
• Infinity fabric allows for the joining of multiple engines on a single die, and offers high bandwidth and low latency. There has been no mention of using Infinity fabric with multiple GPUs.
• Frontier was designed for an array of workload usages. RX Vega is for gaming and will be faster than Frontier.
• Vega will support Tensorflow, Cafe2, Cafe, Torch7 and MxNet via MIOpen.
• Pro versions support hardware virtualization. He did not state out right if this included Frontier/Vega or or not.
• The High Bandwidth Cache Controller (HBCC) helps increase minimum framerates and can improve performance even more if it's specifically coded for.
• Developing drivers for GPUs is really hard.
• Raja expects to grow ROCm to improve machine learning and compute. Another ROCm comment.
• Raja replies to a comment regarding particle physics simulation, saying this will be improved via the new cache and infinity fabric.
• New geometry pipeline in Vega improves throughput per clock cycle and will require no extra work on dev's part to utilize.
• Radeon Vega Frontier will be the fastest single GPU solution for compute.
• Radeon Instinct will provide dramatically better performance per dollar compared to the competition
• RX will have different drivers than Frontier that are optimized for gaming as well as additional goodies.
• Vega is the first GPU architecture to use Infinity Fabric and is in no way a re-hash of Polaris
• Radeon Chill will continue to be improved and will be updated soon.

 

[THESLB]

Estou á espera para ver isso:Main goal of Vega was to create an architecture that can handle large data sets and game at 4K at 60fps

 

[Melli]

Estranho era se não fosse.