AMD Vega Details Leak Ahead Of Official Launch

Status
Not open for further replies.

Nomnom_136

Prominent
Jul 13, 2017
1
0
510
The High bandwidth cache controller is a feature, HBCC, they don't call the memory high bandwidth cache..why are you writing an article about something you know nothing about?
 

mp_4__

Prominent
Jul 6, 2017
1
0
510
How much faster than what? Previous AMD architectures? Current performance king? You need to be a liitle more specific when asking questions. We already know that Vega Frontier Edition is at the level of GTX 1080 so a watercooled Vega would be in between 1080 and 1080 Ti. The big question here must be the pricing : A 1000$ product that cannot beat another at 700 is not even worth considering, so the price of the RX Vega needs to be adjusted according to performance, otherwise will be a big flop.
 

PaulAlcorn

Managing Editor: News and Emerging Technology
Editor
Feb 24, 2015
858
315
19,360


Yes, they do call it high bandwidth cache. From previous coverage.
It’s no secret that Vega makes use of HBM 2; that information was on roadmaps through 2016. But we now know that AMD wants to call this pool of on-package memory—previously the frame buffer—a high-bandwidth cache. Got it? HBM2 equals high-bandwidth cache now. Why? Because AMD says so.

http://www.tomshardware.com/news/amd-vega-architecture-new-features,33265.html


 


I do agree those are the expectations. Although I would like to see Vega reviews first because driver optimizations could easily eek out another 10% performance which would lean more into the 1080ti arena. The key like you said is pricing. Another odd factor is how good this is at mining, well since that is effecting pricing these days, sadly.
 

InvalidError

Titan
Moderator

If the connector got hot enough to melt, then you had a defective connector, crimp or solder job either on the GPU card or PCIe cable. MiniFit Jr pins are rated for 9A each even for the "low spec" variant, which would be enough to safely pass ~300W over the PCIe power cables' three 12V wires.
 

TJ Hooker

Titan
Ambassador

Eh, not necessarily. It's all about hash/dollar and hash/watt. Also, ETH is memory bound, so you aren't pinning the GPU at 100% and hitting those TDPs (can even undervolt/underclock without really affecting performance). In theory ETH should love the bandwidth of HBM2, although it remains to be seen whether the existing algorithms will work with that kind of memory. I know they don't perform particularly well with GDDR5X.
 

DerekA_C

Prominent
Mar 1, 2017
177
0
690
@tech2020 so long as it doesn't hashrate through the roof fury is able to be tweaked to 45 ish hashrate vs the 30 to 32 you can get out of Polaris, if vega pulls 55 ish then there will be a shortage. This has been proven higher hash rate the lower the latency of the memory and that is why hbm is favoriable to such high hash rate we are now talking a second gen of this new low latency tech it will be higher then fury's 45 hashrate unless hbm2 brought a lot of latency but I don't think so or it couldn't be used as a reliable caching system as amd claims to fame with it. So we are looking at at least 50 to 60 hash rate.
 

InvalidError

Titan
Moderator

As far as Ethereum is concerned, Vega won't be anywhere near cost-effective enough to be worth bothering with - not enough RAM to concurrently work on more hashes and put the extra compute to use. Vega FE got benchmarked with Ethereum and performs worse than an R9-390.
 

grozzie

Reputable
Feb 26, 2014
25
0
4,530
Come on! let's not start slagging AMD's GPU's off before they're on the market. I have been waiting for the VEGA GPU since it was announced (ok nearly a year ago). So am looking forward to the desktop version and if you match it with a RYZEN cpu then TDP kind of doesn't matter that much. Anyway if you're really worried about power subsumption why not go for an NVIDIA 1030 - it runs at 35w!!!!!!
 

falchard

Distinguished
Jun 13, 2008
2,360
0
19,790
Right now Vega Frontier can hash at around 40 mh/s and they have not been optimized for the workload yet. Expect them to peak past 50 mh/s on average. If these cards are priced under $500, expect there to be shortages. Especially on the entry level ones as the goal will be to down volt them and use them for their memory.
Purely off the performance of the R9 Fury X and scaling it up with the Vega XTX, it should be competitive with the 1080 Ti after initial driver optimization and consistently better after a year like most AMD cards. This is of course not taking into consideration architecture changes. The question will be how they perform against the Volta GPUs as most expected the Vega GPUs several months ago. Volta will be an architecture change so it will be difficult to tell how it will perform.
 

InvalidError

Titan
Moderator

By the time Vega Xxx becomes available at retail assuming boards hit retail a month after AMD makes the BIOS available to AIBs, Ethereum's working block will have grown by another ~1GB and the hash rate will be cut in half due to 8GB only being large enough to accommodate one Ethereum hash at a time.
 

bit_user

Polypheme
Ambassador

I think it should be good. I believe GDDR5X adds bandwidth partly at the expense of latency. HBM2 should be lower-latency (also, lower-power). But there's also the question of transaction size and channel width.
 

aliveoutofhabit

Reputable
Aug 12, 2015
4
0
4,510
I have a Gigabyte 1080 Turbo with a 180W TDP. Since the XTX water-cooled has 375W TDP does that actually translate into 195W additional required from my PSU? If so... I need to upgrade!
 

FritzEiv

Honorable
Dec 9, 2013
253
0
10,780


Please read this, especially the section on HBM2, especially the section where there's a slide from AMD,calling its own HBM2 "High Bandwidth Cache." And try speaking to other human beings with kindness once in a while.
http://www.tomshardware.co.uk/amd-vega-architecture-new-features,news-54537.html

 

Gillerer

Distinguished
Sep 23, 2013
361
81
18,940


Higher TDP together with higher heat dissipation due to the water cooling will result in the GPU being able to spend longer periods running at higher clocks - even maximum clocks. So yes, it will run faster in situations where the air-cooled variant can't run at max turbo frequency.
 

Cordt_

Commendable
May 10, 2016
6
0
1,510


Actually if you have been paying attention to anything regarding these cards they do refer to HBM2 as HBC which stands for high bandwidth cache. Don't choke on that foot.
 
Status
Not open for further replies.