The launch of Threadripper marks a move closer to AMD’s starting point for the Zen architecture. Contrary to popular belief, AMD did not start its plans with desktop Ryzen and then glue modules together until Epyc was created; no, instead, the company started with an MCM CPU more similar to Epyc, then worked its way down to Ryzen desktop CPUs. Threadripper is the fruition of this MCM design on the HEDT side, and benefits from months of maturation for both the platform and AMD’s support teams. Ryzen was rushed in its weeks leading to launch, which showed in both communication clarity and platform support in the early days. Finally, as things smoothed-over and AMD resolved many of its communication and platform issues, Threadripper became advantaged in its receipt of these improvements.
“Everything we learned with AM4 went into Threadripper,” one of AMD’s representatives told us, and that became clear as we continued to work on the platform. During the test process for Threadripper, work felt considerably more streamlined and remarkably free of the validation issues that had once plagued Ryzen. The fact that we were able to instantly boot to 3200MHz (and 3600MHz) memory gave hope that Threadripper would, in fact, be the benefactor of Ryzen’s learning pains.
Threadripper will ship in three immediate SKUs:
Respectively, these units are targeted at price-points of $1000, $800, and $550, making them direct competitors to Intel’s new Skylake-X family of CPUs. The i9-7900X would be the flagship – for now, anyway – that’s being more heavily challenged by AMD’s Threadripper HEDT CPUs. Today's review looks at the AMD Threadripper 1950X and 1920X CPUs in livestreaming benchmarks, Blender, Premiere, power consumption, temperatures, gaming, and more.
This episode of Ask GN (#56) revisits the topic of AMD's Temperature Control (TCTL) offset on Ryzen CPUs, aiming to help demystify why the company has elected to implement the feature on its consumer-grade CPUs. The topic was resurrected with thanks to Threadripper's imminent launch, just hours away, as the new TR CPUs also include a 27C TCTL offset. Alongside this, we talk Threadripper CPU die layout diagrams and our use of dry erase marker (yes, really), sensationalism and clickbait on YouTube, Peltier coolers, Ivy Bridge, and more.
For a separate update on what's going on behind the scenes, our Patreon backers may be happy to hear that we've just posted an update on the Patreon page. The update discusses major impending changes to our CPU testing procedure, as Threadripper's launch will be the last major CPU we cover for a little while. Well, a few weeks, at least. That'll give us some time to rework our testing for next year, as our methods tend to remain in place for about a year at a time.
Following an initial look at thermal compound spread on AMD’s Threadripper 1950X, we immediately revisited an old, retired discussion: Thermal paste application methods and which one is “best” for a larger IHS. With most of the relatively small CPUs, like the desktop-grade Intel and AMD CPUs, it’s more or less been determined that there’s no real, appreciable difference in application methods. Sure – you might get one degree Centigrade here or there, but the vast majority of users will be just fine with the “blob” method. As long as there’s enough compound, it’ll spread fairly evenly across Intel i3/i5/i7 non-HEDT CPUs and across Ryzen or FX CPUs.
Threadripper feels different: It’s huge, with the top of the IHS measuring at 68x51mm, and significantly wider on one axis. Threadripper also has a unique arrangement of silicon, with four “dies” spread across the substrate. AMD has told us that only two of the dies are active and that it should be the same two on every Threadripper CPU, with the other two being branded “silicon substrate interposers.” Speaking with Der8auer, we believe there may be more to this story than what we’re told. Der8auer is investigating further and will be posting coverage on his own channel as he learns more.
Anyway, we’re interested in how different thermal compound spreading methods may benefit Threadripper specifically. Testing will focus on the “blob” method, X-pattern, parallel lines pattern, Asetek’s stock pattern, and AMD’s recommended five-point pattern. Threadripper’s die layout looks like this, for a visual aid:
Because of the spacing centrally, we are most concerned about covering the two clusters of dies, not the center of the IHS; that said, it’s still a good idea to cover the center as that is where the cooler’s copper density is located and most efficient.
Our video version of this content uses a sheet of Plexiglass to illustrate how compound spreads as it is applied. As we state later in the video, this is a nice, easy mode of visualization, but not really an accurate way to show how the compound spreads when under the real mounting force of a socketed cooler. For that, we later applied the same NZXT Kraken X62 cooler with each method, then took photos to show before/after cooler installation. Thermal testing was also performed. Seeing as AMD has permitted several other outlets to post their thermal results already, we figured we'd add ours to the growing pool of testing.
This week’s hardware news recap goes over some follow-up AMD coverage, closes the storyline on Corsair’s partial acquisition, and talks new products and industry news. We open with AMD RX Vega mining confirmations and talk about the “packs” – AMD’s discount bundling supposed to help get cards into the hands of gamers.
The RX Vega discussion is mostly to confirm an industry rumor: We’ve received reports from contacts at AIB partners that RX Vega will be capable of mining at 70MH/s, which is something around double current RX 580 numbers. This will lead to more limited supply of RX Vega cards, we’d suspect, but AMD’s been trying to plan for this with their “bundle packs” – purchasers can spend an extra $100 to get discounts. Unfortunately, nothing says those discounts must be spent, and an extra $100 isn’t going to stop miners who are used to paying 2x prices, anyway.
Show notes below.
We took time aside at AMD’s Threadripper & Vega event to speak with leading architects and engineers at the company, including Corporate Fellow Mike Mantor. The conversation eventually became one that we figured we’d film, as we delved deeper into discussion on small primitive discarding and methods to cull unnecessary triangles from the pipeline. Some of the discussion is generic – rules and concepts applied to rendering overall – while some gets more specific to Vega’s architecture.
The interview was sparked from talk about Vega’s primitive shader (or “prim shader”), draw-stream binning rasterization (DSBR), and small primitive discarding. We’ve transcribed large portions of the first half below, leaving the rest in video format. GN’s Andrew Coleman used Unreal Engine and Blender to demonstrate key concepts as Mantor explained them, so we’d encourage watching the video to better conceptualize the more abstract elements of the conversation.
Our recent R7 1700 vs. i7-7700K streaming benchmarks came out in favor of the 1700, as the greater core count made it far easier to handle the simultaneous demands of streaming and gameplay without any overclocking or fiddling with process priority. Streaming isn’t the whole story, of course, and there are many situations (i.e. plain old gaming) where speed is a more valuable resource than sheer number of threads, as seen in our original 1700 review.
Today, we’re testing the R7 1700 and i7-7700K at 1440p 144Hz. We know the i7-7700K is a leader in gaming performance from our earlier CPU-bottlenecked 1080p testing; that isn’t the point here. We’ve also pitted these chips against each other in VR testing, where our conclusion was that GPU choice mattered far more, since both CPUs can deliver 90FPS equally well (and were effectively identical). This newest test is less of a competition and more of a “can the 1700 do it too” scenario. The 1700 has features that make it attractive for casual streaming or rendering, but that doesn’t mean customers want to sacrifice smooth 144Hz in pure gaming scenarios. As we explain thoroughly in the below video, there are different uses for different CPUs; it’s not quite as simple as “that one’s better,” and more accurately boils down to “that one’s better for this specific task, provided said task is your biggest focus.” Maybe that’s the R7 1700 for streaming while gaming, maybe that’s the 7700K for gaming -- but what we haven’t tested is if the 1700 can keep up at 144Hz with higher quality settings. We put to test media statements (including our own) that the 1700 should be “better at streaming,” finding that it is. It is now time to put to test the statements that the 7700K is “better at 144Hz” gaming.
This series is an ongoing venture in our follow-up tests to illustrate that, yes, the two CPUs can both exist side-by-side and can be good at different things. There’s no shame in being a leader in one aspect but not the other, and it’s just generally impossible given current manufacturing and engineering limitations, anyway. The 7700K was the challenger in the streaming benchmarks, and today it will be challenged by the inbound R7 1700 for 144Hz gaming.
People like to make things a bloodbath, but just again to remind everyone: This is less of a “versus” scenario and more of a “can they both do it?” scenario.
Under guidelines by AMD that we could show Threadripper CPU installation and cooler installation, we figured it’d also be pertinent to show cooler coverage on TR and RAM clearance. These all fall under the “installation” bucket and normally wouldn’t get attention from us, but Threadripper’s uniquely sized socket with uniquely positioned dies demands more instruction.
Threadripper thermal compound & coldplate coverage has been a primary topic of discussion since we first showed motherboards at Computex. We’ve generally offered that, theoretically, coldplate coverage should be “fine” as long as the two Threadripper CPU dies are adequately covered by the coldplate. In order to determine once and for all whether Asetek coolers will cover the IHS appropriately, seeing as that’s what TR ships with, we mapped out the dies on one of our samples, then compared that to CLC thermal paste silk screens, coldplates, and applied thermal compound.
During press briefings leading to Vega’s gaming variant launch, which looks similar to the FE card (but with DSBR and power saving features now enabled), GamersNexus met with several members of AMD’s RTG team to discuss RX Vega’s future.
One such conversation with a group of media led to the topic of lacking CrossFire marketing materials in RX Vega’s slide decks, with parallels drawn to Polaris’ brandished claims from 2016. With the Polaris launch, great emphasis was placed on dual RX 480 cards evenly embattling GTX 1080 hardware – something we later found to be of mixed virtue. This time, it seems, none of the CrossFire claims were made; in fact, "CrossFire" wasn’t once mentioned during any of the day-long media briefing. It wasn’t until media round-table sessions later in the day that the topic of CrossFire came up.
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.
AMD’s Ryzen lineup mirrors traits at both the R3 and R7 ranges, where both series of CPUs are effectively the same inter-lineup, but with different clock speeds. The R7 CPUs largely all clock to about the same area (+/-200MHz) and consist of the same features. The same can be said for the two R3 SKUs – the R3 1200 and R3 1300X – where the CPUs are functionally identical outside of frequency. This means that, like with the R7 1700, the R3 1200 has potential to challenge and replace the 1300X for users willing to overclock. Remember: A basic overclock on this platform is trivial and something we strongly encourage for our audience. The cost savings are noteworthy when driving an R7 1700 up to 1700X or 1800X levels, and the same can likely be said about the R3 1200.
That’s what we’re finding out today, after all. Our R3 1200 review follows the review of the 1300X and aims to dive into gaming performance, overclocking performance, production applications, and power consumption. Nearby CPUs of note include the 1300X, the Pentium G4560, the R5 series CPUs, and the i3 CPUs.
AMD’s R3 1200 is a $110 part, making it $20 cheaper than the R3 1300X and significantly cheaper than both the i5 and R5 CPUs. Frequency is also down: The R3 1200 clocks at 3.1GHz base / 3.4GHz boost on its 4C/4T design, lower than the R3 1300X that we just reviewed.
We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.