#Bizwhiznetwork.com Innovation ΛＩ |Technology News
2017 has already been a banner year for AMD, with a suite of products more competitive than anything the company could offer since ~2010. Most of the company’s launches, however, have focused on desktop computing. While the desktop market is important to AMD, it doesn’t account for anything like 50 percent of sales — the estimates we’ve seen suggest the desktop/laptop split is more like 30/70 in favor of mobile computing. Any long-term plan to turn AMD’s future and fortunes around, therefore, needs to include a credible mobile solution. Early reviews on the AMD 2500U have surfaced, and we’ve rounded up the results.
First, a bit of explanation. HP is the first company out of the gate with a Ryzen 5 2500U in its x360 15z laptop, and this is Ryzen’s first debut in mobile form factors. This means battery life and even system performance could change with future updates. That’s not guaranteed, but it’s far from unusual, either. Second, with just one point of comparison, this is a specific comparison of one HP model, the HP Envy x360 15z, as opposed to a generalized look at AMD’s Ryzen 5 2500 performance across a range of systems. As we’ve seen in the past with Intel’s Core M and AMD’s Carrizo, the design decisions a laptop manufacturer makes can have a significant impact on final performance.
Hot Hardware and Tech Report have both tested this laptop in a range of applications. We’ll start with gaming, since that’s always been AMD’s strong suit in mobile. The long and short of it is: “Better than Intel, but not faster than Nvidia’s GeForce MX150.” The MX150 is a 384:24:16 (that’s GPU cores, texture units, and ROPs) GPU, while the Vega 8 is supposedly a 512:32:16 GPU (AMD has not confirmed the ROP count). But while AMD has a beefier chip on paper, that GPU has to share its TDP with the CPU, as well as its memory bandwidth — and APUs have always been quite memory bandwidth sensitive.
This graph of Rocket League from Tech Report at 1080p is only one benchmark they ran, but it’s an instructive one. AMD’s Vega 8 outperforms Intel quite well, but it can’t catch the GeForce MX150. Then again, add-on cards in laptops tend to carry a fair price premium, which likely balances out Vega 8’s somewhat lower performance.
CPU performance is a mixed bag. As both Hot Hardware and Tech Report note, applications like Cinebench show the Ryzen 5 2500U performing well against Intel. Handbrake and raytracing are both competitive, and the Ryzen 5 2500U even competes reasonably well in some audio workstation benchmarks. Single-threaded and lightly threaded workloads still tend to favor Intel overall, however, and if you push up to a 45W TDP, Intel’s Core i7-7700HQ sweeps the 15W chips.
Ryzen Mobile performs quite well in rendering and transcoding tests.
Hot Hardware writes that “The early indicators for AMD’s Ryzen Mobile platform are strong, both on the CPU and GPU side of the equation,” while Tech Report states that “Overall, AMD finally has the competitive CPU cores it’s so desperately needed to go with the powerful graphics processors in its APUs, but for the middle of the bell curve, single-threaded performance and snappiness still matters most.”
Our reading of both reviews concurs with this statement. The Ryzen 5 2500U isn’t a knock-out blow for Intel, but it’s simply leagues beyond anything Carrizo or Bristol Ridge could offer. This is the most balanced laptop APU AMD has been able to offer since Llano as far as its competitive position against Intel, and if you wind up trading away some CPU performance, you can generally look forward to best-in-class iGPU performance as a result.
But there is a caveat, albeit an unclear one, to all this.
Battery Life: An Uncertain Situation
The Ryzen 5 2500U’s battery life in the laptops TR and Hot Hardware checked is pretty bad. HH logged the Spectre x360 15z at 272 minutes after installing an updated UEFI from HP that gave them an extra 19 minutes of run-time. TR’s tests also found the x360 15z was similarly underpowered, turning in a runtime of just 6 hours 12 minutes in a web browsing test compared with 11 hours for an Intel system (details available here). Tech Report went so far as to isolate the display entirely and test an external monitor on both an Acer Swift 3 with an MX150 compared with the x360 15z. Battery life with an external display went up ~15 percent for the AMD system (a gain of 53 minutes) and 5 frickin’ hours for the Acer Swift 3.
It’s not clear what’s causing this vast discrepancy, but we’d avoid concluding the 2500U has a major battery life issue just yet. It’s extremely common for mobile systems on new hardware to have battery life issues at launch, to the point that it’s almost the norm. Sometimes the problems are specific to the way a publication tests the machine, sometimes they’re the result of immature drivers and firmware, and sometimes they’re specific to a single manufacturer. Either way, it’s too early to call on this one — a driver update from AMD could improve the situation over time. If, for example, Vega 8’s GPU is running at a higher clock than it should be, that could explain the difference. TR also notes that recent testing of other HP x360 Spectre systems has shown poor battery life — a recent Intel system with a Core i5-8250U actually scored below the AMD system in a head-to-head contest (5hr 49 minutes Intel, 6hr 11 minutes AMD). AMD’s poor battery life showing, in other words, may be entirely HP’s doing.
Our overall takeaway is this: AMD’s 2500U is by far the best Intel competitor the company has launched in years. Performance is only one issue that laptop customers buy on, but if AMD can bring its APUs to market below Intel’s quad-core 8th generation mobile CPU prices, it’ll have a compelling price/performance argument for itself for the first time in a long time. Battery life will play a part in just how attractive these systems are, but hopefully further driver and firmware updates can improve these metrics. Either way, we’d wait to see what other vendors can do with the APU before concluding Ryzen 5 has weak battery life.
