AMD's Ryzen CPUs (Ryzen/TR/Epyc) & Vega/Polaris/Navi GPUs

I hope it is not true as well. If it is though there may be allot of used cards on the market. About the only save grace may be these will eat power like they need their own feed from the pole.

 

I don't think mining systems already built will be abandoned, at least until they payout their costs, and even then it's probably too much trouble to take them down if they are showing a profit.

Especially with the power and cooling demands so high with the new RX Vega 64 (and 56 OC), new systems would likely need to be built.

Buy the boards where and when you find them, even with a bumped up price, if nothing else you can flip it if you don't want to keep it

 

Hell, if these cards are on the higher end of that estimate (near 100MH/s), then 4-8 cards would pay my basic living needs (if not counting recoup). TBH, that wouldn't be too bad. I'd also check out other alt-coin to try to diversify my holdings as the boom in eth is over (but would see a slight rise). If I free up those costs, plus have what earned otherwise, then...

 

8 cards, it would provide the home heating as well........

 

AMD Threadripper Round-Up!

ULTIMATE Ryzen 7 Build Guide: Episode 1!

How to Build an INSANE Ryzen Gaming & Editing PC!: Episode 2!

 

Good news and bad news, new Vega FE tweaks boost performance 34%... for Monero Mining...

RX Vega Amazing At Mining??

 

Alienware Area-51 THREADRIPPER EDITION


First AMD Ryzen Threadripper 1950X Benchmarks Published
http://wccftech.com/first-amd-ryzen-threadripper-benchmarks/

 

Primitive Discarding in Vega: AMD Mike Mantor Interview

 

AMD RX Vega x2 Allegedly In The Works By Asus, To Be Liquid Cooled
http://wccftech.com/amd-vega-64-x2/

"A liquid cooled dual GPU Radeon RX Vega graphics card is allegedly in development by Asus and should grace store shelves later this year. We’ve heard whispers that design work is underway at Asus to put two of AMD’s upcoming high-end Vega 10 on a single board."

RX Vega X2 GPU Rumored - Fastest Single Card GPU ? Ryzen & Linux Issues


ASUS Reportedly Building Radeon RX Vega 64 X2 Dual GPU Beast
https://hothardware.com/news/asus-building-radeon-rx-vega-64-x2-dual-gpu

The ASUS Ares II was a dual HD 7970 graphics card and a powerhouse for its time

 

2X Vega GPU's on a single card, would that be 4 8 pin power plugs to the card?

 

I think when (if?...) it's full technology suite works and at the right voltage/clocks, vega will be quite efficient. I think the power thirst of FE (and it also seems the 64) is because they are specced way beyond that point of... dare I call it... maxQ? *runs*

Bad news about its mining hash. Unless you want a dirt cheap used gpu setup...

 

Vega is like a completely failure outside of mining. AMD need to figure out whether it wants to make a computing card or make a gaming card. Even if AMD make a good computing card, it doesnt have the software infrastructure for it anyways. It needs to completely strip out any computing features and make a lean gaming card like Pascal.


Also using HBM2 on consumer platform is a terrible idea, it reduced power consumption but it likely delayed the GPU and increases cost quite a bit.

 

Tbh, I think if it is done by Asus, it's a research joint venture and will implement IF potentially between the dies. Asus is good at dual chip cards, without doubt. But, if they are given permission for an exclusive dual gpu for a limited period of time in exchange, that would interest me more than anything.

Sent from my SM-G900P using Tapatalk

 

Ryzen 3, The Ultimate Gaming Benchmark Guide!

 

It's already been confirmed that the power saving features were disabled on the Vega FE, and drivers weren't optimized for gaming, or anything else.

Performance on the RX cards and FE will improve after release as well, so just like the RX 480 and Ryzen early release, there is a lot to look forward to in the long run, and even at release strong 1070 / 1080 contention and low price, plus the advantage of not being Nvidia will sell lots of cards.

Hopefully there will be enough cards to go around to gamers and miners.

 

​

Someone claims they found LinusTechTips 1950x TimeSpy (using the Dellware Alien desktop) & gave the link: http://www.3dmark.com/spy/2152841

 

That processor ID is the AW engineering sample that has been out thee for a while., not a production unit.

 

Might I ask on what are you basing this assertion?
RX Vega is still not officially released, and there are unconfirmed benchmarks about Vega 56 (cheapest of the lot) being close to 1080 in performance and much far ahead of 1070 for a lower price.

Furthermore, most of Vega's actual features have been confirmed to not being used in the first place on Vega FE due to drivers (which leads me to my next question: 'did anyone try to benchmark Vega FE gaming performance with latest Crimson drivers?) - this includes power saving mode during gaming, rasterization, etc.

Also, I don't think that Nvidia stripped out any compute features out of its Pascal cards. Nvidia artificially reduced compute performance via software, as is evident through their new driver release that increased compute performance by about 3 times on Titan (barely matching or slightly exceeding Vega's, but still Vega FE managed to pull ahead in many benchmarks representative of real life workloads).

This is an issue with the ridiculous 'pro' market segment which really needs to be shut down, and both AMD and Nvidia should just focus on making consumer grade GPU's that can do both and alternate between different modes via software (ala Vega FE).
Realistically, there are no intricate hardware differences between consumer and pro cards (except perhaps VRAM size)... main differences are in drivers.

Perhaps. HBM here is a bit of an oddity as its supposed to drop power consumption over regular GDDR5, but AMD made a choice to create Vega on a manuf. process that's suitable for low clock speeds - hence why it might end up consuming more power once its clocked beyond certain point - that and overvolting the core to increase yields.
AMD chose to push boundaries with a newer technology that would be brought to everyone (consumers) so people could eventually benefit from it.

Do I think they might have opted for a different strategy such as perhaps create Vega with regular GDDR5 and HBM2 for example?
Yes.
That could have likely kept production issues with Vega at a minimum, but AMD also pushing for HBM might actually force developers to start optimizing games to make use of this massive bandwidth.

Besides pushing for proprietary software, Nvidia didn't really opt to bring something like this to others.

I can't see Nvidia 'innnovating' like AMD does.
Even their proprietary hairworks software (which uses tesselation) is really nowhere near TressFX (AMD's open source software for hair) which works better on either hardware.

 

You'd be better off running a nuclear power plant at that point.

 

Nvidia doesnt artificially decrease computing performance via driver. The quadro have similar Fp16/64 performance to geforce. What happened on that driver was something else. I have a titan x pascal and I tested it myself. Essentially, no computing performance were increased per say, it bought much higher performance via quadro style optimization for certain pro software. But if you were using FP16, you really shouldnt get any increases there....

Pro lines need to be exist. FE drivers arent promised the same reliability and ECC as Firepro drivers. To be honest also, when you are paying 50k+ a year for commercial licenses, you dont care about the quadro cost.Frankly, no one really complains about the quadro costs that uses them because those people are too busy complaining about the licensing costs...

Bringing out a new technology at very high expense and have very little benefit isnt innovation, its throwing money away. AMD cant make both gddr5x and hbm2 work on the same die just like nvidia. They should have never used Gddr5x. The power saving helps but.... Vega is such a disaster performance per watt front anyways... AMD's HBM2 also didnt even push the bandwidth well.... 2 stacks is compared with 384bit bus with gddr5x anyways...

As for Vega 56, what are the chances that it will be stay at MSRP and not be inflated due to mining?

Whether you look at performance per watt or performance per die size, AMD is 50%+ behind nvidia and that is what matters.

Essentially what nvidia did with the computing side it is they split their Tesla dies and their Geforce dies. So before we had GK110 on both Geforce cards and Tesla cards. The Geforce card had unnecessary computing features on them that arent being used and were wasted. With the new line, nvidia had all of the vital computing HPC features on their GP100 and had a separate consumer die of GP102. This is more efficient and resulted in higher performance on both side.

In short, RTG dont have the budget nvidia does and are falling behind significantly.

 

RTG indeed doesn't have Nvidia's budget... and that may slow things down, but did we ever stop to think what exactly is their focus with Vega?
Seems to me that AMD is trying to get a foot in the door in multiple areas with their products, and even if gaming performance on Vega is a bit lower than anticipated (that it may not go up against 1080ti - which remains to be seen), it would still be high enough to reclaim their position in the high end while using their superior compute capability in other areas (mining, AI, etc.).

As for performance per watt - as I said before, AMD overvolts their GPU's frequently to increase yields, and is using a manuf. process suitable for lower clock speeds, not higher ones. Past a certain point, you get diminishing returns... - heck even Nvidia isn't experiencing massive performance increases due to overclocking their GPU's.
Look at RX 580. The silicon is of obviously higher quality and the GPU runs at higher clocks than reference RX 480 while using lower voltage. OEM's seem to like to push RX 580 past its usable limits on better silicon resulting in higher power draw.
Apparently, they didn't stop to think that offering say same or slightly higher clocked version of RX 480 while using FAR LOWER voltages (and subsequently power draw) might make Polaris look even more appealing to some people.

Also, do you remember Fury Nano?
And now also the upcoming Vega Nano?
Both seem to show a very low TDP due to binning, lower voltage and probably slightly lower clock speeds (with Fury Nano, performance down only 10% compared to their entry level Fury - a minor performance loss for a massive drop in power consumption - and subsequent testing showed that undervolting brought down Fury X power consumption very close to Maxwell while maintaining performance - Nvidia can afford to optimize their GPU's power consumption out the door, AMD can't - it's a trade-off for now).

Vega FE when undervolted showed the architecture can even beat P5000 and P6000 quadro's at pro workloads with early drivers for a fraction of the price.

If its' capable of doing that on PRO software, chances are its gaming capability might also be able to accomplish the same (but this remains to be seen).
Plus, AMD generally packs more hardware onto it's gpu's in general making it more power hungry... but it also tends to report its overall power consumption vs the average one.

 

Their computing probably ONLY gets used on mining due to a lack of ecosystem in anything else.

As for PRO software performance, it really depends. I do not remembering seeing a FE beating a P6000 in the specperfwise. You also have be aware of different renders used and etc. It really doesnt translate to gaming performance at all.

AMD is also very late this time. Late on the market and I doubt even RX nano can have the performance per watt of GTX 1080. Vega is really really not that good for gaming.
Just get a GTX 1080 for the same price and call it a day.

And the whole marketing blind test, dont even get me started on that....

 

Please stop trying to sell Nvidia here, this is an AMD thread, I for one would rather buy an AMD GPU and pair it with an AMD CPU.

No sense feeding Nvidia or Intel, I'd rather vote with my $ for AMD.

The AMD CPU performance is well ahead of Intel, and the AMD GPU performance close enough even at launch to build new desktop and hopefully new laptops with AMD only.

 

Ok? I see companies all the same, it doesnt matter what different names they have or colors, their loyalty is to their shareholders and their shareholders only. That means profiting off the consumers. I simply get the best product. The colors or names of the companies dont concerns me. Ofcourse, you are welcomed to your own opinion.

 

The shady practices and lack of innovation of Intel, and the failures of Nvidia are getting old, time to fund AMD and get them strong again so there is competition in the market.

Already Nvidia unlocked features of the Titan Xp allowing pro features and 3x-5x better performance because AMD released the Vega FE.

Intel is already trying to make up for lost time to catch up to AMD CPU offerings.

Both Intel and Nvidia have been milking us for a long time, and AMD have provided us a chance to stop that cycle.

Pay attention to the company practices, not their "colors", sheeesh!!

 

I pay for the best product. Because to me as a individual, what matters for me its get the best product.
AMD makes better CPUs, so my next build I am likely using Threadripper. If AMD wants my money, it needs to provide a product thats better than nvidia for my purpose first. Otherwise, I dont give charities to public for profit companies. Ofcourse, this is purely my opinion and I think it would best if we do not discuss this anymore.

 

AMD is the best product hardware and software performance wise in many categories now, especially when you consider the companies selling the alternatives.

AMD GPU's are performing better than the 1060, 1070, and 1080, so for 95% of the people looking for a GPU the AMD solution will give them what they need performance-wise, and give AMD the funding to keep building better hardware and maintain competition in the CPU and GPU markets.

 

Last time AMD had better CPU Intel cheated with anti competitive crap with OEMs. They got fined but the damage was already done. Hopefully this does not happen again.


Sent from my iPhone using Tapatalk

 

Dirty tricks aren't usually seen while they are happening, dirty tricks will be found out and exposed much later, long after the damage has been done. Here are some good history and commentaries, there are a lot more available:

Intel - Anti-Competitive, Anti-Consumer, Anti-Technology.


Are we MAD at Intel?!


Nvidia, stop being a DICK


Nvidia GameWorks - Game Over for You.


Why Are AMD & NVIDIA So Polarizing?


At this point I find value and comfort in supporting alternatives whenever possible.

 

HW News - RX Vega Mining, Partner Cards (...)

 

If the leaked benches are accurate, the cheapest Vega 56 will exceed 1070 while selling for less money than Founder Edition and approach 1080 level of performance at 210W TDP (bear in mind that Vega 56 can probably be undervolted to radically reduce power consumption to around 1080 level or less, eliminate potential thermal throttling and allow the boost clocks to be maintained for as long as one needs them - if thermal throttling presents a problem - also, AMD reports total power draw, not average one - we know Nvidia GPU's do go above their advertised TDP).

In this particular instance Vega 56 for example becomes an attractive option.
I don't particularly care if AMD is late to the party. We just have to deal with the fact they are the underdog and given their financial status, its amazing what they were able to release on the CPU front.

Also, we don't know really if Vega architecture is a bust.
You are basing your opinions on Vega FE and jumping the gun.
If the driver on Vega FE was able to raise mining performance to the level it has, why is it impossible that it can do similar/same for gaming?
We also know that AMD's drivers sometime need extra time to mature better in order for performance to be fully extracted.. about 3 to 6 months, though RTG were quite consistent with fast driver releases anyway, and are actually better than Nvidia at this point.

'Best product' is relative, considering that Nvidia merely shrunk Maxwell and overclocked it to get Pascal.
They also paid extensively to get their proprietary software and hardware to be pushed as opposed to using open standards like AMD does.
Yes, AMD is also a corporation, however, they have a consistent track record in supporting open standards even when things were going for them, Intel and Nvidia do not.

If you want to support Intel and Nvidia that do have a clear record of anti-competitive practices, then its your decision.
I will reserve my judgment of RX Vega and its capabilities until it's fully released and evaluated.

 

FreeSync vs. G-Sync in 2017, Informational Video!

 

Spoiler: Twitter Feud Brewing?

Old Frank Azor (Dellware's GM) sounds hot n' bothered by our unsuspecting hero Linus and his Alien GTFOOH smackdown review. PDF Setup manual which they've pulled back off their support site (before word gets out???) says their lack-luster Alienboard is max 2933MHz DDR4 rated, or as Linus calls it 'an unexceptional motherboard with limited overclocking' while slumming with unsightly 2667 RAM. What else is new. Hmmmm. To deflect, Frank hints at new memory of unknown speed at unknown date. Ambiguity sounds about right. I mean, 2933 is faster than 2667 ... final specs need to be revisited at later date, since faster memory is on the way (!). Spoiler below is best suited to the 'hmph' folder for now

Spoiler: Preliminary Max Mem Speed 2933

early adopters paying premium surcharge for budget 2667 that's on offer now (AlienTax)?

​

 

Some other news pertaining to RX Vega:
http://www.gamersnexus.net/news-pc/3004-rx-vega-64-and-vega-56-power-specs-price

"We confirmed power targets independently at 165W for the RX Vega 56 model and 220W for the RX Vega 64 model, with total board power for the RX Vega 56 estimated at 210W on the reference model, and about 290W for the RX Vega 64 air reference model. The Aqua model has a total board power of 350W. One note here: AMD changed these numbers depending on who spoke to, so there’s a chance other outlets will report different numbers of 150W and 220W.

As for power tuning, we asked whether this was just tightening of the power target or if actual power performance features were enabled under the hood. Turns out, RX Vega isn’t just a matter of restricting power target, they’re actually doing something for power optimization. We couldn’t get explicit examples at this time. One thing we do know is that the voltage targets change, so voltage checks are at different frequencies than FE, and voltage should be lower. We’d expect that this will align with our findings in the undervolting testing on Vega: Frontier Edition, where power consumption can equalize while improving performance. It’s still AVFS, but just a better tuning profile than FE.

We also asked AMD’s architects, including Mike Mantor, about whether DSBR was actually disabled in Vega: Frontier Edition or whether it was just a rumor. The architects loosely confirmed that tile-based rasterization was in fact disabled for Frontier Edition’s launch, which we think mostly aligns with statements about pushing the card out in time, and noted that DSBR will be enabled on both Vega FE and RX Vega on launch of RX Vega. We asked about expected performance or power consumption improvements, but were not given any specifics at this time. Wait for launch on that, though – we’ll get that information."

Lower TDP than expected for Vega 56 and 64.
I was under the impression that Vega 56 will be at 210W and 64 at 300W as most websites seem to put them on.
Were the numbers revised in the meantime or something else?

 

​

Old Frank's seein' Ripper Red in his eye's. He's hot to trot now, boi. On tweeting at him why he won't put a like on Linus' review, or retweet it on his & Dellware's twitter page, there's what he said. Looks like he got the paid shills at notHotHardware to cover for him on their dinky platform, some damage control, gave those fellas a Monday 'exclusive' write-up with Benchmarks & dialogue to help fend off some hate. Man, if I could personally engineer that Psychological Game called: let's you n' him fight ... ... ...

We can be sure that Frank & Lisa Su had nothing but the highest regard for Linus last week; how else did he get all the exclusive earlybird perks & peaks & unveils ... (cause of his accurate adequate reviews on a gynormous platform)

​

 

"They love you when you're on all the covers
When you're not then they love another"

-Marilyn Manson
The Dope show

 

AMD LET ME BUILD THEIR NEW 80 VEGA SUPERCOMPUTER…

 

First 4 Things To Do After Ryzen Build

 

Link;
http://www.dsogaming.com/news/threa...unced/attachment/bykski-ryzen-threadripper-6/

 

New Threadripper X399 (Asrock) Motherboard & More Unboxing Goodness!

 

Turns out, the AW can hit 3000 with the slower ram. The issue was Dell sent pre-production silicon to reviewers.

Sent from my SM-G900P using Tapatalk

Edit: remember those Intel payments long ago? Makes you wonder why they sent out review units that only gave 2800-2900 on cinebench with pre-production silicon. Maybe to make Intel look closer in score?

 

Read also my old post
Suddenly increased more performance, Yees

 

Immediately after that, AMD previewed a beta driver privately to show the hash rate could be so much more. One person supposedly used it on the FE and got 70MH/s on a card not really made for that use. AMD has been saying they will unlock features in the future when ready on a couple products, including EPYC. Looks like they were not lying, but are trying to get them stable. Nvidia, though, held it back to do like Intel--up sell the consumer.

 

Isn't this same **** practiced by cable companies? I am sure comcast, verizon, etc can get much better bandwidth but they are locked down on purpose.


Sent from my iPhone using Tapatalk

 

Yes and no. The no part has to do with our infrastructure and US backbones for the internet, as well as the lack of development for it to support the traffic and higher speeds. The other part is lack of competition.

Sent from my SM-G900P using Tapatalk

 

THIS!!

 

Makes sense. When I was in Korea three years ago, basic cable was 100/100 mb


Sent from my iPhone using Tapatalk

 

More companies do than don't.

 

The US infrastructure is getting worse and worse, just so corporations can make more. If we kept the rate increases on taxing oil, we would pay $0.25-0.45 more per gallon, but could repair our entire roadways. We also need more people with the Army Corp of Engineers to inspect our failing dams and bridges. The reason we don't have more is they also perform some of the needed environmental impact studies for things like the Dakota Access Pipeline, which if the study isn't completed in the 90 day window (IIRC, that is the time for them and the BLM and MMS, which was renamed after the gulf oil blowout), they get approved using similar info from nearby locations that were studied. By not funding them, nor staffing them, corporations get permission at the cost of our citizens. It is disgusting what greed can do. I could go on and on about our infrastructure!

 

Full line-up of i9 processors. Look at those prices! Better off going for Threadripper and the extra PCIe lanes.

http://cdn.mos.cms.futurecdn.net/xDvMfKTw8NHLhsh2TCipHW.jpg