We've got a new thermal paste applicator tool that'll help ensure consistent, equal spread of TIM across cooler surfaces for future tests. As we continue to iterate on "Hybrid" DIY builds, or even just re-use coolers for testing, we're also working to control for all reasonable variables in the test process. Our active ambient monitoring with thermocouple readers was the first step of that, and ensures that even minute (resolution 0.1C) fluctuations in ambient are accounted for in the results. Today, we're adding a new tool to the arsenal. This is a production tool used in Asetek's factory, and is deployed to apply that perfect circle of TIM that comes pre-applied to all the liquid cooler coldplates. By using the same application method on our end (rather than a tube of compound), we eliminate the chance of users changing application methods and eliminate the chance of applying too much or too little compound. These tools ensure exactly the same TIM spread each time, and mean that we can further eliminate variables in testing. That's especially important for regression testing.
This isn't something you use for home use, it is for production and test use. When cooling manufacturers often fight over half a degree of temperature advantage, it would be unfair to the products to not account for TIM application, which could easily create a 0.5C temperature swing. For consumers, that's irrelevant -- but we're showing a stack of products in direct head-to-head comparisons, and that needs to be an accurate stack.
The Titan X (Pascal) DIY “Hybrid” project has come to a close, and with that, we've reached our results phase. This project has yielded the most visible swings in clock performance that we've yet seen from a liquid cooling mod, and has revealed significant thermal throttling in the reference nVidia Titan XP design. What's more, this card will not feature the market saturation created by AIB partners with lower end cards, and so more advanced coolers do not seem to be available without going open loop or DIY.
Our liquid-cooled Titan X Pascal Hybrid has increased the card's non-overclocked frequency by an average of nearly 200MHz – again, pre-overclock – because we've removed the thermal throttle point. The card has also improved its clock-rate stability versus temperature and time, provable during our two-hour endurance run.
We've just finished testing the result of this build, and the results are equal parts exciting and intriguing – but that will be published following this content. We're still crunching data and making charts for part 3.
In the meantime, the tear-down of our reader's loaner Titan X (Pascal) GPU has resulted in relatively easy assembly with an EVGA Hybrid kit liquid cooler. The mounting points on the Titan XP are identical to a GTX 1080, and components can be used between the two cards almost completely interchangeably. The hole distance on the Titan XP is the same as the GTX 1080, which is the same as the 980 Ti, 1070, and very similar to the GTX 1060 (which has a different base plate).
Here's the new video of the Titan X build, if you missed it:
With thanks to GamersNexus viewer Sam, we were able to procure a loaner Titan X (Pascal) graphics card whilst visiting London. We were there for nVidia's GTX 10 Series laptop unveil anyway, and without being sampled the Titan X, this proved the best chance at getting hands-on.
The Titan X (Pascal) GP102-400 GPU runs warmer than the GTX 1080's GP104-400 chip, as we'll show in benchmarks in Part 3 of this series, but still shows promise as a fairly capable overclocker. We've already managed +175MHz offsets from core with the stock cooler, but want to improve clock-rate stability over time and versus thermals. The easiest way to do that – as we've found with the 1080 Hybrid, 1060 Hybrid, and 480 Hybrid – is to put the card under water cooling (or propylene glycol, anyway).
In this first part of our DIY Titan XP “Hybrid” build log, we'll tear-down the card to its bones and look at the PCB, cooling solution, and potential problem points for the liquid cooling build.
Here's the video, though separate notes and photos are below:
MSI has begun filling-in its X99A line of Broadwell-E motherboards with workstation-targeted options, built for compliance with ECC Registered DIMMs and with boosted maximum data throughput via M.2. The motherboard fits LGA2011-3 socketed CPUs, including Haswell-E and Broadwell-E, and supports SLI with nVidia Quadro GPUs for production workloads. Additional focus is placed on storage controllers and HSIO allocation, fitting for a board that will be deployed in workstation environments (e.g. render machines, CAD/ProE machines).
The X99A Workstation motherboard uses what appears to be an 8-phase power design for its core VRM, with additional phases for the memory. The VRM is comprised of titanium inductors with a max temperature of 220C, supporting higher current for extreme overclocks. Dark capacitors (solid caps) populate the board and VRM's capacitor bank, rated for a 10-year lifespan.
It looks like that 4K screenshot we were provided was created in some magical developer environment. No Man's Sky is the least stable game that we've worked on in years. Most games with poor optimization are still playable, and are consistent in their pitfalls; consistency permits some level of comparative benchmarking. With No Man's Sky, we're seeing nearly constant stutters and spectacular frame latency spikes in excess of 4000ms. The game exhibits severe stuttering that makes it unplayable at times and, by extension, impossible to accurately benchmark.
We've generated a few sets of benchmark data on a Titan X Pascal, 980 Ti, and RX 480 specifically to demonstrate just how wildly unpredictable the performance is. This is not an instance where we can just test anyway and produce charts as normal, because the FPS range is so wide that you'd end up with performance results that make no sense – like a GTX 1080 performing equally to a 980 Ti in averages in one test, but the opposite in another. That's due to variance introduced from somewhat unpredictable frame latency fluctuations, something we explain in part in this video.
Game-specific GPU benchmarks serve a single purpose: Hierarchically ranking the “best” products for each graphics configuration, resolution, and budget. The very heart of game benchmarking is to produce an objective comparative analysis between components. We have decided to present our findings with No Man's Sky, but have opted out of an immediate graphics card benchmark. This is because, in our eyes, such a benchmark would not be fair to the GPUs. The variance in results is so great that listings end up chaotic, and so we end up constrained and benchmarking the poor performance of No Man's Sky, rather than the performance of the cards themselves. The game is inadequate as a test platform, and cannot be trusted to generate reliable, replicable data from one test iteration to the next.
This GPU performance analysis of No Man's Sky looks at stutters and frame drops (“hitching”), poor optimization, screen flickering, and low FPS.
We recruited Libor “Buildzoid” Sadilek of Actually Hardcore Overclocking to assist in our latest coverage of AMD's RX 460 GPUs. The full review of the Sapphire RX 460 Nitro is located here, with a tear-down of the card over here. Today, we're focusing on the electrical component quality of the Sapphire RX 460 Nitro VRM, along with PCB quality in general.
The Sapphire RX 460 Nitro uses an overpowered VRM, but the cost of the end product is not necessarily offset by this. We'll see if prices stabilize as stock becomes more prevalent, though. NVidia and AMD have both been selling out of stock in short order with their new architectures.
This coverage is entirely video driven. You can find the video embedded below, but be sure to subscribe to the YouTube channel for future “specials” like this one.
Following the Sapphire RX 460 Nitro 4GB graphics card review that we posted, we decided to send the card through a tear-down, as we did with the RX 470, RX 480, GTX 1060, and GTX 1080 (links go to disassembly articles).
The RX 460 Nitro uses a custom PCB and shroud. This is a step away from the reference coolers provided by AMD for the RX 470 and RX 480 cards. The Nitro is easily dismantled, done by removing a handful of rear-side screws to release the shroud & cooler, then four more screws to release the heatsink.
Let's run through some photos and discussion of the PCB. Here's a video of the process, for more perspective:
AMD's fanfare surrounding CrossFire with the RX 480s demanded a test of the configuration, and we decided to run the architecturally similar RX 470s through the same ringer. We only have two RX 470s presently in the lab, and they're not the same card – but we'll talk about how that impacts testing in a moment. The cards used are the Sapphire RX 470 Platinum Edition ($200) and the MSI RX 470 Gaming X, tested mostly in DirectX 11 and OpenGL titles, with some DirectX 12 Explicit Multi-GPU testing toward the end.
The benchmark runs a performance analysis of two CrossFire RX 470s versus a single RX 470, single RX 480, CrossFire RX 480s, and the latest GTX cards (1070, 1060). We're looking at framerate and CrossFire power draw here, with no thermal testing. Read our RX 470 review for in-depth thermal and frequency stability analysis (and overclocking).
We liked the RX 470 well enough, which, for our site, is certainly considerable praise; we tend to stick just with the numbers and leave most of the decision-making to the reader, but the RX 470 did receive some additional analysis. As we stated in the review, the RX 470 makes good sense as a card priced around $180, but not more than that. That's the key point: Our entire analysis was written on the assumption of a $180 video card, presently fielded only by PowerColor and its Red Devil RX 470. Exceeding the $180 mark on a 4GB 470 immediately invalidates the card, as it enter competition with AMD's own RX 480 4GB model (see: 4GB vs. 8GB VRAM benchmark). Granted, it's still far enough away from the RX 480 8GB & GTX 1060 that the 470 may exist in some isolation. For now, anyway.
But as seems to be the trend with both nVidia and AMD for this generation of graphics cards, the RX 470 has some pricing that at times seems almost silly. Take, for instance, the $220 XFX RX 470 RS Black Edition True OC card: it's $20 more than a 4GB RX 480, it's clocked to where we overclocked on our RX 470, and it will perform about 3-5% slower in AVG FPS than the RX 480 4GB reference card. And let's not start on the seemingly irrelevant $240 8GB RX 470 Nitro+, effectively an RX 480 8GB card (even in clock-rate) with four fewer CUs, fewer TMUs (from 144 to 128), and slower memory – though it does have a better cooling solution, to Sapphire's point.
We moderate comments on a ~24~48 hour cycle. There will be some delay after submitting a comment.