The prices are $400 for the RX Vega 56, $500 for the RX Vega 64, and we think $600 for the liquid-cooled RX Vega 64 Aqua. AMD’s launching these with different bundles for their other products as well, but we’ll talk about that momentarily. Today, we’re providing details on the RX Vega specifications, pricing, and other preliminary information (like TDP/TGP) for the GPU. We’ll have a separate content piece coming out shortly that provides a deeper dive on the Vega GPU architecture.

The RX Vega 64 flagship launches at $500 for the reference card – and so likely the range is $500 to $600 for AIB partner models, which would include your standard Strix, Twin Frozr, Windforce, and other coolers. Liquid-cooled models will clock higher by way of reduced power leakage, as we previously showed, though air cooled models can also accomplish this to some lesser but non-trivial extent. AMD’s liquid-cooled model did not carry a standalone price, but had a bundle price of $700 for the card with various discounts for other parts. More on that later.

There’s no doubt that most the news circulating right now will pertain to AMD’s new driver update – and it’s an impressive update, one which we’ll discuss below, but we wanted to revive the “gaming” & “pro” mode discussion.

In speaking with AMD about its “Gaming” and “Pro” toggle switch in the Vega drivers – something we previously demonstrated to be a UI-only switch – we learned that the company intends to do something more meaningful going forward. As of now, the toggle is nothing more than a psychological switch, limiting its usefulness to removing the WattMan button from the UI – not all that useful, in other words. Functionally pointless for Vega: FE as it launched, and symptomatic of a driver package which was either woefully incomplete or intended to encourage a placebo effect.

Ask GN returns for its 54th episode – we’ve gotten more consistent than ever – to discuss Noctua fan manufacturing locations (China & Taiwan), thermal pads vs. thermal paste usage on MOSFETs, Vega 10-bit support, and a couple other items.

A few of the items from this week peer into GN’s behind-the-scenes workings, as several viewers and readers have been curious about our staff, whether we keep products, or why we “waste” GPUs by using them for things other than mining.

As always, timestamps below the embed.

This feature benchmark dives into one of the top requests we received from our Patreon backers: Undervolt Vega: Frontier Edition and determine its peak power/performance configuration. The test roped us in immediately, yielding performance uplift largely across the board from preliminary settings tuning. As we dug deeper, once past all the anomalous software issues, we managed to improve Vega: FE Air’s power available to the core, reduce power consumption relative to this, and improve performance in non-trivial ways.

Although power target and core voltage are somewhat tied at the hip, both being tools for overclocking, they don’t govern one another. Power target offset dictates how much additional power budget we’re willing to provide the GPU core (from the power supply) in order to stabilize its clock. GPU Vcore governs the voltage supplied, and will generally range from 900 to 1250mv on Vega: FE cards.

Vega’s native DPM configuration runs its final three states at 1440MHz, 1528MHz, and 1600MHz for the P-states, with DPM7 at 1600MHz/1200mv. This configuration is unsustainable in stock settings, as the core is both power-starved and thermally throttled (we’ll show this in a moment). The thermal limiter on Vega: FE is ~85C, at which point the power and clock will fluctuate hard to try and maintain control of the core temperature. The result is (1) spikey frequencies and frametime latencies, worsening perceived performance, and (2) reduced overall performance as frequency struggles to maintain even 1528MHz (let alone the advertised 1600MHz). To resolve for the thermal issue, we can either configure a more intelligent fan curve than AMD’s stock configuration or create a Hybrid card; unfortunately, we’re still left with a new problem – a power limit.

The power limit can be resolved in large part by offsetting power target by +50%. Making this modification is easy and “fixes” the issue of clock-dropping, but introduces (1) new thermal issues – resolvable by configuring a higher fan RPM, of course, and (2) absurdly high power consumption for a non-linear scaling in performance. In order to truly get value out of this approach, undervolting seems the next appropriate measure. AMD’s native core voltage is far higher than necessary for the card to operate at its 1600MHz target, and so lowering voltage improves performance from the out-of-box config. This is for thermal and power reasons alike. We ultimately see significantly reduced power consumption, to the tune of ~90W in some cases, a more stable core clock and thereby higher performance, and lower temperature – and thereby controllable noise.

We can’t get all the way down to the inner workings of the pump on this one, unfortunately, as all of our source images for the Vega: Frontier Edition – Watercooled card are from a reader. The reader was kind enough to remove the shroud from their new WC version of Vega: FE so that we could get an understanding of the basics, leading us to the conclusion that AMD has built one of the most expensive pre-built liquid cooling solutions for a graphics card.

The video tear-down goes into detail on the images we received, but we’ll revisit most of it here. The card uses the same base PCB, same VRM, same GPU/HBM layout and positioning, and same everything as the air-cooled card. The difference is entirely in the cooling solution, where the Delta VRM fan goes away and is replaced with an additional reservoir (more on that in a moment), while the GPU/VRM cooling is handled by liquid plates and a pump. The die-case finstack atop the I/O is also now gone, and the baseplate is simplified to an aluminum plate with no protrusions.

Liquid-cooling the AMD Vega: Frontier Edition card has proven an educational experience for us, yielding new information about power leakage and solidifying beliefs of a power wall. We also learned that overclocking without thermal barriers (or thermal-induced power barriers) grants significant performance uplift in some scenarios, including gaming and production, though is done at the cost of ~33A from the PSU over 12V PSU power.

Our results for the AMD Vega: Frontier Edition liquid-cooling hybrid mod are in, and this review covers the overclocking scalability, power limits, thermal change, and more.

The Hybrid mod was detailed in build log form over in part 1 of the endeavor. This mod wasn’t as straight-forward as most, seeing as we didn’t have any 64x64mm brackets for securing the liquid cooler to the card. Drilling through an Intel mounting plate for an Asetek cooler, we were ultimately able to get an Asetek 570LC onto the card, which we later equipped with a Gentle Typhoon 120mm fan. VRM FET cooling was handled by aluminum finstacks secured by thermal adhesive, cooled with 1-2x Corsair ML120 fans. That said, this VRM cooling solution also wasn’t necessary – we could have operated with just the fans, and did at one point operate with just the heatsinks (and indirect airflow).

Before getting started: Our Vega FE Hybrid mod has just gone through its final data pass, and is now in video editing and writing. The content will arrive tomorrow!

That cleared away, as we know a lot of folks are excited for the mod's results, we're now focusing on the MSI GTX 1080 Ti Lightning card momentarily. This is a video card that we first covered at Computex 2017, where we detailed initial specifications, MOSFETs and power components, and the target use case of XOC or heavy overclocking. We didn't yet have information on the card internals, but our latest tear-down (embedded below) gives some insight on the card's design. There are some unique features on this card that should pose an interesting A/B test during thermal benchmarking.

Zotac's GTX 1080 Ti AMP! Extreme is one of the largest GTX 1080 Ti cards on the market, rivaling the Gigabyte Aorus Xtreme card in form factor. The card uses nearly three expansion slots, runs a long PCB and cooler, and hosts a dense aluminum heatsink with a three-fan cooler. This card runs $750 to $770, depending on if the “Core” edition is purchased. The only difference is the out-of-box clock, but all these 1080 Tis perform mostly the same in games (once solving for thermals).

For its VRM, Zotac takes a brute-force approach to the 1080 Ti, using a doubled-up 8-phase (16 phases total) with rebranded QN3107 and QN3103 MOSFETs, operating on a UP9511 in 8-phase mode. The VRM is the reason for the tall card, with two phases tucked off to the side (under a small aluminum heatsink that's isolated from all other cooling). This theoretically helps distribute the heat load better across a larger surface area, which Zotac then cools using a small aluminum fin stack that's isolated from the denser aluminum fin array. Above the VRM's isolated heatsink rests a rubber damper, which doesn't fully make contact (and is presumably to prevent scratching in the event of over-flex during installation, as it otherwise does nothing), and then the three fans.

zotac amp extreme contact 1

Above: Contactless rubber bumper above the MOSFET heatsink.

zotac 1080ti extreme 2

The card is one of the heaviest, largest cards we've looked at this generation. To give some perspective, Zotac's AMP Extreme is about 1” thicker than a 2-slot card (like the reference card), is longer than the Aorus Xtreme, and is heavy from the mass of aluminum resting atop the GPU. Learn more about the inner-workings of this card in our tear-down.

For today, we're focusing on thermals, power, and noise, as that's the biggest difference between any of these 1080 Ti cards. The gaming performance and overclocking performance, sans Kingpin/Lightning cards, is not notably different.

We’ve already endured one launch of questionable competence this quarter, looking at X299 and Intel’s KBL-X series, and we nearly escaped Q2 without another. Vega: Frontier Edition has its ups and downs – many of which we’ll discuss in a feature piece next week – but we’re still learning about its quirks. “Gaming Mode” and “Pro Mode” toggling is one of those quirks; leading into this article, it was our understanding – from both AMD representatives and from AMD marketing – that the switch would hold a relevant impact on performance. For this reason, we benchmarked for our review in the “appropriate” mode for each test: Professional applications used pro mode, like SPECviewperf and Blender. Gaming applications used, well, gaming mode. Easy enough, and we figured that was a necessary methodological step to ensure data accuracy to the card’s best abilities.

Turns out, there wasn’t much point.

A quick note, here: The immediate difference when switching to “Gaming Mode” is that WattMan, with all its bugginess, becomes available. Pro Mode does not support WattMan, though you can still overclock through third-party tools – and probably should, anyway, seeing as WattMan presently downclocks memory to Fury X speeds, as it seems to have some leftover code from the Fury X drivers.

That’s the big difference. Aside from WattMan, Gaming Mode technically also offers AMD Chill, something that Pro Mode doesn’t offer a button to use. Other than these interface changes, the implicit, hidden change would be an impact to gaming or to production performance.

Let’s briefly get into that.

Reader and viewer requests piled high after our Vega: Frontier Edition review, so we pulled the most popular one from the stack to benchmark. In today’s feature benchmark, we’re testing Vega: FE vs. the R9 Fury X at equal core clocks, resulting in clock-for-clock testing that could be loosely referred to as an “IPC” test – that’s not exactly the most correct phrasing, but does most quickly convey the intent of the endeavor. We’ll use the phrase “academic exercise” a few times in this piece, as it’s difficult to draw strong conclusions to other Vega products from this test; ultimately, GPUs simply have too many moving parts to simulate easier IPC benchmarks like you’d find on a CPU. As one limitation is resolved, another emerges – and they’re likely different on each architecture.

Regardless, we’re testing the two GPUs clock for clock to see how Vega: FE responds with the Fury X in the ring.

We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.

Advertisement:

  VigLink badge