AMD Ryzen 9 4900HS Overview

AMD Ryzen 9 4900HS Overview

Cell computing has change into the following large goal at AMD with its new sequence of Ryzen 4000 APUs. The brand new processors will arrive because the U-series for low-power and ultraportables and the H-series for prime efficiency laptops. At this time we now have the primary retail Ryzen 4000 laptop computer available on the market to judge, so the main focus will probably be on the benchmarks, so we are able to see for the primary time how Zen 2 introduced throughout to cellular stacks up in opposition to Intel’s tried and true line-up.

All Ryzen 4000 APUs are primarily based on the Zen 2 structure, constructed utilizing 7nm expertise with a monolithic die that additionally incorporates a Vega-based GPU. Earlier this month we printed a full breakdown of Ryzen 4000 APUs, detailing a few of its architectural options and specs. We’ll skip most of these particulars on this assessment, however it is best to undoubtedly examine that out for an summary of what else is coming in 2020.

Simply to be clear, Ryzen 4000 components don’t use AMD’s Zen 3 structure. The naming is just a little complicated right here – Ryzen 3000 for Zen 2 desktop chips, and Ryzen 4000 for Zen 2 APUs – however that is merely a continuation of what AMD began in 2017 once they launched the primary Zen APUs as Ryzen 2000. Ryzen 4000 APUs are thus primarily based on Zen 2 cores, similar as the most recent Ryzen desktop CPUs.

It is no coincidence Ryzen Cell U and H sequence mirror what Intel gives for several types of laptops. With ultraportables launch nonetheless pending, at this time is all in regards to the highly effective H-series and the way AMD will fare for content material creation, productiveness and gaming.

AMD’s Ryzen 4000 H-series consists of three distinct SKUs: the Ryzen 5 4600H delivers 6 cores and 12 threads for the mainstream, the Ryzen 7 4800H bumps that as much as 8 cores and 16 threads, alongside a higher-clocked 8-core variant within the Ryzen 9 4900H. All of those APUs include a 45W TDP, and are complemented by a low energy 35W variant for laptops that additionally adhere to the remainder of AMD’s HS-series design tips.

You’ll additionally discover totally different GPU configurations for every SKU, utilizing the refreshed Vega design with as much as 8 compute models and better clock speeds. Fascinating to notice, however we’d anticipate most H-series laptops to additionally embody a discrete GPU, like our take a look at system. Cache, whereas listed at 12 MB, is definitely a mixed L2 and L3 determine: all components have 8 MB of L3 cache, Ryzen 7 and 9 get 4MB L2, and Ryzen 5 will get 3MB.

The chip we’re benchmarking at this time is the Ryzen 9 4900HS, a 35W SKU that gives barely increased base and enhance clock speeds than the Ryzen 7 4800H that sits beneath it at 45W, a little bit of binning magic at play there. It’s not the outright flagship APU in AMD’s line-up, however it should give us a extremely stable have a look at Ryzen 4000 efficiency. We’ll additionally offer you a sneak peak of Ryzen 7 4800H efficiency from an engineering pattern, as we glance to assessment the remainder of processors within the line-up as new laptops make it to market.

Our benchmark testbed is the Asus Zephyrus G14, a neat transportable 14-inch gaming laptop computer that packs in Ryzen 4000 HS APUs and as much as GeForce RTX 2060 Max-Q graphics. We obtained the highest-spec mannequin for testing, with the RTX 2060 Max-Q and the 4900HS, 16 GB of DDR4-3200 reminiscence and a 1080p 120Hz show.

The deal with this assessment is only CPU efficiency, so we can’t go into the design and different options of the Zephyrus G14. We’ll say this although, after per week of use we expect it’s a really properly constructed laptop computer, with a fantastic keyboard and as we’re about to see, very compelling efficiency in such a small kind issue.

You’ll additionally see throughout this assessment quite a lot of totally different laptop computer CPUs within the charts and, at occasions, GPU configurations as properly. The information within the charts is a mean from the laptops we’ve examined with the given {hardware}. Testing laptop computer elements is of course a bit tougher than desktops as every configuration can differ in cooling and different {hardware}, so these averages are supposed to illustrate how a ‘typical’ system will carry out. The averages don’t embody single channel reminiscence programs or another conditions that closely throttle the elements at hand, we’ve tried our greatest to create apples-to-apples knowledge the place doable.


Let’s kick issues off with a traditional efficiency benchmark: Cinebench R20. On this take a look at we see whole domination by the crimson crew. Not solely is the 4900HS the quickest laptop computer processor we’ve examined within the multi-threaded workload, it’s additionally the quickest in single-thread efficiency. The 4900HS is 35% quicker than the 8-core Core i9-9880H when utilizing all cores, and seven% quicker in single-core. There may be the Core i9-9980HK that we haven’t examined that may put up a little bit of a combat, however the 90W numbers from the 9880H recommend this is likely to be a tough activity. Additionally think about, these outcomes are from the lower-TDP 4900HS, the 4900H at a full 45W must be one other step forward.

There are different brutal outcomes on this Cinebench chart. The 4900HS crushes the Core i7-9750H, which is the most well-liked CPU utilized in slim and light-weight gaming notebooks. Zen 2 providing is over 60% quicker within the MT take a look at. We additionally see the previous-gen AMD half get a bit humiliated, the 4900HS is a lot quicker than the Ryzen 7 3750H in the identical energy envelope it’s not humorous.

We additionally examined with the legacy Cinebench R15 which has outcomes from a wider vary of CPUs going again extra years. The 4900HS stays firmly positioned on the high of the charts as anticipated. Anybody upgrading from a Core i7-7700HQ, for instance, which was a preferred H-series CPU in 2017 will probably be handled to greater than double the efficiency with a Ryzen 9 4900HS.

One space the place Zen 2 receives a very massive improve is in the way it handles AVX-256 directions. It’s merely a lot quicker at broad floating level operations than earlier than. So the Ryzen 9 4900HS receives an infinite efficiency improve in our Handbrake x265 take a look at, which makes use of AVX directions. The 4900HS is a monstrous 179% quicker than the 3750H on this workload, which is simply unfathomable for a single technology efficiency leap.

On an AMD vs Intel entrance, it’s additionally extremely favorable for AMD. Not fairly the identical margins as we noticed in Cinebench, however we nonetheless see a 23% efficiency benefit of the 4900HS over the 9880H, and a forty five% benefit over the 9750H. Frankly, these are huge deltas for a laptop computer kind issue that usually receives single digit gen-on-gen enhancements.

For Intel to match the 4900HS with its 8-core providing, it must blow its energy goal out the window and as an alternative use a 90W TDP, which is feasible on some gaming laptops with a ‘Turbo’ mode or related. The distinction in energy draw on the wall for these two programs is unimaginable: the G14 with the 4900HS ran comfortably at round 66W long run, in comparison with 150W for the facility boosted 9880H in our HP Omen 15 take a look at system. That simply goes to indicate how way more environment friendly AMD’s Zen 2 design is at these long run workloads.

Blender tells us the same story, with big efficiency beneficial properties within the ~35% vary for the 4900HS over the 9880H, and ~65% vary for the 4900HS over the 9750H. Most Blender customers will in all probability render on the GPU as an alternative because it tends to be a lot quicker, nevertheless it’s one other benchmark that illustrates long run multi-core efficiency on these laptops.

By way of decompression, we see AMD draw back with a big victory in 7-Zip. Ryzen processors are recognized to work very properly on this workload. Nevertheless it does fall behind the 9880H by way of compression, to the tune of ~8%, though AMD beats decrease core rely components just like the 9750H.

Adobe Photoshop efficiency is attention-grabbing. In our Iris Blur take a look at which is usually CPU restricted, the 4900HS manages to beat the 9750H however falls behind the 9880H, sitting comfortably between these two processors.

This continues to be the case within the extra complete Puget Photoshop benchmark, which runs via a spread of exams. In each of those workloads, which apply numerous results to very excessive decision photos, the 4900HS is round 10 p.c slower than the 9880H in a core-for-core battle. Granted, we’re seeing 40% increased efficiency than AMD’s last-gen APUs.

Now let’s take a look at PCMark 10 numbers. We’re wanting on the Necessities and Productiveness workloads as they’re CPU restricted, whereas the remainder of the exams rely extra on the GPU.

The Necessities workload covers issues like app loading, internet shopping and video conferencing. AMD manages to match the efficiency of the 9880H, which can not sound all that spectacular, till you understand that the last-gen 3750H will get slaughtered on this take a look at. Now AMD is at efficiency parity for these on a regular basis workloads.

On the Productiveness workload we see one thing related as earlier than, with the Ryzen 9 4900HS outperforming one thing just like the Core i7-9750H. These types of duties had been lower than first rate on earlier Ryzen, so though Ryzen 4000 isn’t crushing Intel right here, efficiency parity is an effective end result.

Nevertheless there are some conditions with productiveness workloads the place Ryzen 4000 isn’t as spectacular. Our customized Excel benchmark options plenty of quantity crunching on a big dataset, and right here that the 9880H outperforms the 4900HS once more. The 4900HS is quicker than the 9750H, so it’s not a horrible end result by any means, nevertheless it does appear that giant dataset workloads are a weak spot for Ryzen.

MATLAB lets us reconfirm this with our ODE and FFT benchmarks. Once more, a lot of knowledge crunching on massive units of knowledge and Intel’s 9880H walks away with a victory.

One other workload the place Ryzen performs properly, however not sufficient to beat Intel’s 8-core competitor is our Acrobat PDF export take a look at, which is absolutely single-threaded. The 9880H is marginally quicker in Acrobat, though we nonetheless see a big efficiency improve for the 4900HS over the 3750H with a wholesome 20% bump to single-threaded efficiency.

One closing workload earlier than we have a look at GPU acceleration is AES-256 efficiency as supplied by SiSoft’s Sandra benchmark. Within the multi-threaded take a look at, we see 15% increased AES efficiency from the 4900HS versus 9880H to make it the quickest CPU we’ve checked out for cryptography workloads. This offers Ryzen a neat benefit in two closely utilized low-level duties in decompression and cryptography.

Now let’s work via our Premiere exams that are principally GPU accelerated, beginning with our 1 cross encode which takes benefit of Intel’s QuickSync expertise. Premiere doesn’t help {hardware} accelerated encoding on AMD processors at this stage, so for those who like a fast export with barely lowered picture high quality on the finish, Intel continues to be the best way to go. Particularly, the 4900HS paired with Nvidia’s GeForce RTX 2060 Max-Q is ~20% slower than a Core i7-9750H with an RTX 2060 just because {hardware} acceleration just isn’t out there.

Alternatively, our Core i9-9880H system doesn’t help QuickSync acceleration because the iGPU is absolutely disabled to help G-Sync via the Nvidia GPU. So we’re left with a pleasant software program encoding comparability between the 9880H and 4900HS, the place the 4900HS pulls away strongly to the tune of 26%.

Then we get to our 2 cross encode which produces superior picture high quality and doesn’t help {hardware} acceleration. On this state of affairs, efficiency falls again to what we noticed from most long-term workloads: the 4900HS is 30% quicker than the 9880H, and 38% quicker than the 9750H, regardless of our 9880H laptop computer packing a a lot quicker GPU. This take a look at isn’t GPU restricted however even with a reasonably succesful discrete GPU it seems as if AMD’s Zen 2 APU is considerably quicker for Premiere encoding.

However wait, there’s extra Premiere exams. Right here we now have a single occasion of Warp Stabilizer, which is a vastly demanding impact that runs on a single thread per occasion. The 4900HS manages to stabilize the footage 14% quicker than the 9880H, and 22% quicker than the 9750H, which reveals the facility of the excessive single-thread efficiency we first noticed with Cinebench.

And eventually we now have the Puget benchmarks. The Ryzen 9 4900HS is one of the best laptop computer CPU for stay playback in Premiere, performing 12% higher than the 9880H, that means we’re seeing higher enhancing efficiency. Then for his or her export benchmark, we see related numbers to what we’ve simply been speaking about, with QuickSync acceleration helping with a few of this workload.

Gaming Benchmarks

Whereas we don’t suppose many individuals will truly be utilizing the built-in GPU in these H-series processors given the overwhelming majority of H-series laptops additionally embody discrete graphics, it’s value a short have a look at how the iGPU fares in a small number of low depth video games, simply to see what enhancements AMD has made to the Vega GPU.

The Ryzen 9 4900HS sports activities the quickest iGPU configuration out there within the sequence with 8 compute models clocked as much as 1,750 MHz. And it does ship spectacular outcomes. In Grand Theft Auto V, the 4900HS delivered 36% extra efficiency than the Ryzen 7 3750H, which featured 10 Vega compute models at 1,400 MHz. On paper, each CPUs have related uncooked GPU efficiency, however with all the benefits AMD talked about like elevated reminiscence bandwidth, this new Zen 2 APU is ready to pull forward.

In Civilization VI, the 4900HS once more offers 37% extra efficiency than the 3750H inside the similar 35W energy envelope. And shifting to CS: Go, that margin stays equivalent at round 37%. The 4900HS operating CS Go completely on the built-in graphics was in a position to obtain over 100 FPS on common utilizing low settings, which may be very good.

After which lastly we now have our most performance-intensive iGPU take a look at in Gears 5 operating at Medium settings. The 4900HS is round 31% quicker than the 3750H on this workload, which makes it quicker than a low-end discrete GPU providing like Nvidia’s MX250.

It’s not frequent to see an H-series laptop computer paired with an MX-class GPU. Often OEMs go for extra highly effective configurations like a GTX 1650 or increased, however with Ryzen 4000 there’s actually no must trouble with an MX250 or related discrete GPU in an H-series sort design.

Sustained Clocks and Extra Questions

At this level within the assessment we’d usually stroll you thru some comparability summaries between the Ryzen 9 4900HS and numerous different CPUs. However we nonetheless suppose there’s just a few efficiency questions left on the desk. One is what clock speeds the CPU truly runs at in follow, and the way enhance behaves. And the opposite is, why can we see decrease efficiency in some knowledge heavy workloads like Matlab, Excel and Photoshop. So let’s deal with this second half first.

Now we have two working theories as to why we see knowledge heavy workloads carry out this fashion. The primary is an easy one primarily based on the specs of those processors, specifically cache sizes. The Core i9-9880H has an honest 16 MB of L3 cache, which matches what Intel supply on the desktop with components just like the i9-9900K. Nevertheless the Ryzen 9 4900HS has simply 8 MB of L3 cache, half of what Intel gives, and properly beneath the 32 MB of L3 that AMD packs into their 8-core Zen 2 desktop processors just like the Ryzen 7 3700X.

Having much less cache means much less knowledge may be saved in tremendous quick reminiscence and accessed right away. When you have got excessive core rely, excessive efficiency CPU cores however not sufficient cache, this will change into a bottleneck in some cases. And whereas this cache quantity has doubled on Ryzen 4000 from earlier cellular sequence, so has the core rely. It’s in all probability not the entire story, however undoubtedly a part of it.

The opposite is the reminiscence system. Sure, AMD gives increased reminiscence bandwidth than Intel with the transfer to help DDR4-3200 speeds, Intel solely gives DDR4-2666 with Ninth-gen. In a benchmark like Sandra we see AMD offering round 35% extra reminiscence bandwidth. Nevertheless, Ryzen 4000 seems to have inferior reminiscence latency. This, like cache measurement can change into a efficiency constraint. The Core i9-9880H has reminiscence latency round 30ns for knowledge units above 32MB in measurement, whereas the Ryzen 9 4900HS has 46ns reminiscence latency. That’s a considerable win for Intel.

It’s arduous to say for positive whether or not these elements are a part of the trigger, or your entire trigger, however wanting via low stage benchmarks these had been the 2 issues that stood out to us. Once we get to benchmark extra Ryzen 4000 APUs we’ll get a clearer image of the place these bottlenecks lie.

As for clock speeds and enhance conduct, let’s have a look. From a chilly begin in our Handbrake AVX workload, the Ryzen 9 4900HS constantly boosted as much as round 65W of energy for just a few seconds attaining 4 GHz enhance clocks all core, earlier than dropping right down to 54W for an extended sustained interval, with clocks round 3.7 GHz. Finally the CPU settles right down to 35W to offer 3.2 GHz all-core, simply above this processor’s 3.0 GHz base clock. The enhance interval can differ relying on how heat the system is, however from a chilly begin we had been typically seeing at the least 2 minutes of ~53W enhance which is beneficiant. Temperatures had been truly very properly managed on the Zephyrus G14, with its air cooler offering a tick over 70C sustained, however after all, this may differ between laptops.

Efficiency Abstract

Ryzen 9 4900HS vs. Core i9-9880H

The large one right here is the Ryzen 9 4900HS versus the Core i9-9880H. These are related class processors, 8-core versus 8-core and may give us a fairly stable indication of what to anticipate from Tenth-gen as properly, if we get minor clock pace enhancements from Intel as is rumored.

In most long run, multi-threaded workloads, we’re seeing 35% higher efficiency from the Zen 2 chip. We additionally get spectacular numbers for video encoding. Single-threaded workloads are additionally usually quicker, or within the worst case eventualities like gentle productiveness, equal to Intel’s. Nevertheless, efficiency does fall behind in knowledge heavy benchmarks like Excel, Photoshop and Matlab.

The wonderful factor about most of these outcomes is AMD is ready to obtain, in lots of workloads, at the least 25% higher multi-thread efficiency whereas additionally coming in with decrease sustained energy draw, at 35W versus 45W. So what occurs whenever you restrict Intel’s 9880H to a PL1 of simply 35W utilizing Intel’s XTU software program?

Effectively, the margins develop much more. Whereas we do let these processors do what they usually do within the enhance stage, some benchmarks don’t change an excessive amount of, long run workloads now are strongly in favor of AMD. Over 50% higher efficiency in some conditions is feasible, and a 40% enchancment to Handbrake is spectacular. Which means an OEM designing a slim and light-weight system with restricted cooling will typically get one of the best expertise by far going with Ryzen.

Amazingly, even once we throw Intel’s 45W energy restrict out the window and enhance as much as 90W perpetually, Intel nonetheless can’t beat the 35W AMD processor in long run multi-thread workloads. In an astounding win for effectivity, the 4900HS is both equal to this 90W processor configuration, or as much as 15% quicker. With these two programs going at it, we noticed an 80W energy draw distinction from the wall, so if Intel needs to match AMD with 14nm CPUs by simply elevating energy limits, laptops will want a lot bigger coolers to manage.

Ryzen 9 4900HS vs. Core i7-9750H

Evaluating the 4900HS to the 9750H, there’s just one occasion the place the 9750H system may be quicker, and that’s with QuickSync accelerated Premiere encoding. In each different benchmark, together with these the place beforehand the 9880H was quicker, now the 4900HS is quicker. This bodes properly for any future battles within the Ryzen 7/Core i7 vary the place AMD has a core benefit in providing completely 8-core components.

Ryzen 9 4900HS vs. Ryzen 7 3750H

Now let’s have a look at how far AMD has include Zen 2 vs. Zen+ in a 35W cellular design. The distinction in sure workloads is staggering. Not solely is AMD producing a 30% efficiency enchancment in single-thread workloads, they’re in a position to ship 2.5x the efficiency in long run multi-thread duties like Blender, Handbrake and Cinebench — with the identical energy draw.

Ryzen 9 4900HS vs. Ryzen 7 4800H

We did promise a sneak peek at Ryzen 7 4800H efficiency on an engineering pattern laptop computer we had temporary entry to. Happening a small subset of benchmarks, efficiency is wanting equal to, if not barely higher, than the 4900HS, which bodes very properly for extra mid-range programs that can use the Ryzen 7 half. Sure, the 4900HS is extra environment friendly, however the 4800H ought to nonetheless ship pure 8-core efficiency.

What We Realized

Ryzen 4000 is delivering one of the best efficiency for heavy productiveness workloads you may get in a cellular kind issue. We will say that with out even testing the upper energy Ryzen 9 4900H, as a result of the 35W variant within the Ryzen 9 4900HS is already leaving Intel’s 8-core rivals within the mud.

The Ryzen 9 4900HS is a superb alternative for core-heavy work: we’re speaking video encoding or transcoding, 3D rendering, file compression, that type of factor. It’s a lot quicker and does so at a decrease energy draw, permitting for higher efficiency in a smaller kind issue. Even in single-thread workloads, AMD typically comes out on high, which implies that the Premiere enhancing expertise is best on the Ryzen 9 4900HS, for instance.

From a pure gen-to-gen improve perspective, AMD has performed a tremendous job. The Ryzen 7 3750H was troublesome to suggest and it was handily crushed by Intel’s mid-range components. That has all modified with the newer APUs.

Technically, Intel components can nonetheless compete on the efficiency entrance in the event you take away the 45W long run energy restrict. We noticed that with the Core i9-9880H and we anticipate it might be related had we examined the flagship Core i9-9980HK. However to get that type of efficiency, you want a bigger laptop computer with a beefier cooler, and it’ll doubtless run hotter, too. The Ryzen 9 4900HS can present all that efficiency in smaller laptops — like this Zephyrus G14 — and run cooler on the similar time.

Nevertheless, AMD’s Ryzen 4000 CPUs aren’t the whole bundle, there are just a few limitations. One is with knowledge heavy workloads, like quantity crunching huge spreadsheets in Excel, working with big pictures in Photoshop or operating Matlab scripts. Intel got here out forward in these exams. Lighter hundreds will do exactly high-quality on AMD, however in the event you’re an enormous quantity cruncher, Intel’s core-equivalent processors is likely to be the best way to go.

The opposite is with workloads that use Intel unique applied sciences like QuickSync, Premiere encoding being one. If that’s one thing you utilize on a regular basis, Intel is the best way to go. It’s value mentioning that each different activity in Premiere outdoors of QuickSync encoding is quicker on Ryzen 4000, together with enhancing and making use of intensive results just like the Warp Stabilizer.

Outdoors of these two cases, you’ll have a greater, quicker productiveness expertise with these Ryzen 4000 8-core processors. That’s uncharted territory for AMD on a cellular PC kind issue.

Additionally it appears that evidently AMD will probably be providing this type of efficiency at very aggressive costs. The Zephyrus G14 with the Ryzen 9 4900HS and RTX 2060 Max-Q is about to retail for about $1,500, which is firmly the territory of six-core Core i7-9750H laptops. The Core i9-9880H laptop computer we purchased for testing on this assessment value $2,400 and was one of many least costly we might discover. Granted, it does have a a lot quicker GPU for gaming, however proper now you simply gained’t discover 8-core laptops within the $1,500 worth vary. Whereas most of our knowledge has been specializing in Ryzen 9 vs Core i9, they don’t actually compete face to face on pricing, at the least for now as this might change with Intel’s Tenth-gen.

Just a few further notes to wrap up this assessment…

With earlier Ryzen laptops we didn’t have essentially the most secure expertise, operating into loads of software program bugs and crashes, which we have mentioned just a few occasions earlier than. Utterly totally different expertise with the Ryzen 9 4900HS within the Asus Zephyrus G14. Even with a beta model of the GPU driver on this laptop computer, we bumped into no stability points, crashes or functions refusing to work appropriately.

You additionally is likely to be questioning about gaming efficiency with a discrete GPU. The Zephyrus G14 does embody the RTX 2060 Max-Q in any case. That’s one thing we’ll discover in a future assessment as we’re nonetheless working to get one of the best apples to apples comparability. Hoping to have that quickly.

Battery life will even be on individuals’s minds, nevertheless it’s not one thing we got down to take a look at as we did not have sufficient knowledge for numerous laptops to make it a good comparability.

Lastly, we’ll shut this one out with some feedback on Intel’s Tenth-gen. The brand new laptop computer CPUs are coming simply across the nook, doubtless inside a month. Rumors recommend Intel is sticking with 8-core components for the H-series, and are nonetheless utilizing 14nm expertise. We simply can’t see how they will compete on a efficiency or effectivity stage. If all we’re getting is a small clock pace bump with the identical 14nm effectivity as the previous few generations, Ryzen goes to simply win this one. We’ll see that battle unfold quickly, however primarily based on what we have seen up to now, we don’t maintain excessive hopes for Tenth-gen.

Buying Shortcuts

Leave a Reply

Your email address will not be published. Required fields are marked *