AMD’s second-generation Ryzen Mobile APUs go Pro

Have you tried a laptop with one of AMD's recent Ryzen Mobile APUs? They're actually pretty great. My buddy picked up a machine with a Ryzen 5 3500U and after putting an SSD in it (because, sigh, it came with only a hard drive), it's super speedy. If you'd like to deploy a whole bunch of similar systems to your corporate workforce, don't worry—AMD's just released "Pro" versions of some models from that series. Check 'em out:

2nd-generation

Ryzen Pro Mobile APUs

TDP Cores /

Threads

CPU clock

(Base / Boost)

GPU core

configuration

GPU peak

clock

Process

Node

Ryzen 7 Pro 3700U 15W 4C / 8T 2.3 / 4.0 GHz 10 CUs (640 SP) 1.4 GHz 12nm
Ryzen 5 Pro 3500U 15W 4C / 8T 2.1 / 3.7 GHz 8 CUs (512 SP) 1.2 GHz 12nm
Ryzen 3 Pro 3300U 15W 4C / 4T 2.1 / 3.5 GHz 6 CUs (384 SP) 1.2 GHz 12nm
Athlon Pro 300U 15W 2C / 4T 2.4 / 3.3 GHz 3 CUs (192 SP) 1 GHz 14nm?

Only four of the second-generation Ryzen Mobile chips—just the 15 W models—made it to the Pro lineup, and unsurprisingly, the core configurations, clock rates, and basically everything else remain the same as the non-Pro chips. One of the chips that made the grade is the Athlon 300U, now christened the Athlon Pro 300U. It's interesting to see AMD offering such a low-end processor in "Pro" form given that to get Intel's similar vPro technology you'll have to step all the way up to a Core i5.

As we noted back when these same chips launched in non-Pro format, the Ryzen 7 3700U's ability to pack four two-thread Zen+ cores and ten Vega compute units into a 15W package that can hit 4 GHz is pretty darn impressive. AMD pits that chip in benchmarks against Intel's Core i7-8650U and claims that its processor comes out ahead in various tests, including Cinebench, Photoshop, PCMark, and 3DMark.

If you're not familiar with AMD's Pro lineup, you can refer to some of our earlier coverage to get caught up. The short version is that these processors offer management and security features that aren't available in the regular desktop and laptop Ryzen chips. AMD's Pro-series chips have been a little difficult to find in the past, but the chipmaker has a whole page on its site that lists available machines mounting its microprocessors.

Of course, none of the systems on that site are equipped with the latest chips listed above. AMD says it expects "commercial systems from HP and Lenovo" later this quarter with other OEMs and "further platform updates" coming in the second half of the year.

Comments closed
    • tipoo
    • 5 months ago

    Now that they threw a “Pro” suffix in there, how about these for an update to the nontouchbar Pro, eh Tim Apple?

    • gerryg
    • 5 months ago

    Would the Athlon Pro 300U be used in education or non-profit settings maybe? I could see penny-pinching companies and government usage being possible use cases. Maybe for cheap all-in-ones? Just brainstorming.

      • K-L-Waster
      • 5 months ago

      Entry level laptops would seem to be a natural fit.

        • LocalCitizen
        • 5 months ago

        Pro tho?

          • derFunkenstein
          • 5 months ago

          In some environments. This is perfect for schools who want to have 1:1 device programs with Windows machines rather than Chromebooks or iPads, for example.

    • ronch
    • 5 months ago

    These Ryzen Mobile chips may not be the fastest but Zen is actually pretty nifty in low power applications. Take for example the 2500U. If you’re in the market for a gaming laptop you’ll probably have to put the 2500U alongside the Core i5 8300H if you’re shopping for something around $700-$800. The 8300H beats the Ryzen by about 25% (inversely the Ryzen delivers about 80% the performance of the i5) but the 2500U is a 15w TDP chip while the 8300H is 35w. That’s especially critical in a computer with limited cooling capabilities and that has to run off the battery while on the go. And oh the 2500U’s power budget takes into account the much more potent Vega graphics.

    For a 4C/8T CPU with a potent GPU and 15w TDP, Ryzen Mobile is pretty cool. I’m not really buying the Pro moniker but I doubt Jim and Dwight care either.

      • cynan
      • 5 months ago

      Except that to run at full clock speeds, they tend to require a bit more juice than what they’re rated (an AMD-infamous TDP tactic).

      Even worse, these chips end up being shoehorned in ultrabooks with mediocre cooling designed to handle something like 12W TDP max, with very little room for adaptive voltage algorithms to do any good whatsoever, particularly on the GPU. I speak from experience with an HP Elitebook with a 2700U that never reaches close to full advertised potential on GPU clocks.

        • Goty
        • 5 months ago

        [quote<]Except that to run at full clock speeds, they tend to require a bit more juice than what they're rated (an AMD-infamous TDP tactic). [/quote<] 1) TDP is not the same as power consumption, and hasn't been for quite a while. (Technically it never [i<]was[/i<] but it was close enough as made no difference.) 2) Intel does exactly the same things with its parts, in both mobile and desktop. See the following for some examples and explanation as to why: [url<]https://www.anandtech.com/show/13400/intel-9th-gen-core-i9-9900k-i7-9700k-i5-9600k-review/21[/url<]

          • cynan
          • 5 months ago

          The distinction between TDP and power consumption is moot.

          Point being that I agree that RyZen mobile are great chips, but AMD, relative to Intel, just does not seem to be able to get OEMs to bother designing proper implementations. With RyZen mobile, this is seen with single-channel memory configs hamstringing Vega 10, and inadequate cooling (with OEMs such as HP simply undervolting the GPU in the bios instead). Intel/Nvidiadoes fall prey to these issues from time to time, but not anywhere near the rate of AMD.

          • DavidC1
          • 5 months ago

          TDP is defined in Intel datasheets as [i<]sustained average power consumption[/i<], and older models(including desktops) have followed that pretty well. It can exceed that TDP, but it has to come back down to meet the sustained Pavg. With many year delays of 14nm and continued competitive pressures, Intel has deviated from the definition and are using sometimes sleazy tactics to pretend the older metrics are wrong. Well, in desktops they can pull this nonsense, but in laptops and thermally constrained setups they'll be in trouble. Lot of 15W U chip based systems are already using the 25W cTDPup mode as well. You still need to dissipate that heat, otherwise you'll have long term problems.

            • ronch
            • 5 months ago

            That’s right. 35w Intel CPUs like the i5-8300H seem to be having thermal throttling issues these days. I reckon 35w has long been a ‘standard’ mobile CPU TDP; it’s well-known how to cool a 35w chip. But if 35w is suddenly getting hot you know something isn’t right. You might accuse AMD of going a little bit past 15w at times but Intel’s guilty of it too, and at the end of the day 15w is still way below 35w for 80% the performance.

      • Goty
      • 5 months ago

      Isn’t the 2500U TDP configurable? I thought it was possible to run it at 25W. If so, I wouldn’t be surprised to see it running in that configuration in less thermally-constrained systems like the ones you’re talking about.

        • ronch
        • 5 months ago

        It’s hard to say whether a laptop goes beyond the nominal TDP but I’d wager most mainstream gaming laptops wouldn’t unless it’s a pricier laptop in which there’s more leeway for the OEM to be able to afford spending more money on cooling. In my experience though, my Acer Nitro 5 with a 2500U + discrete Radeon 560X runs very quietly while the competing 8300H is known for heat and throttling issues, even in other laptops like the Dell G3.

        • Spunjji
        • 4 months ago

        Yes; technically. My understanding is that the processor if configured as a”standard” will draw down up to 25W for turbo boost / active GPU scenarios if the APU is being cooled sufficiently. Good luck finding out how it’s configured in any given design, though. 🙁

      • psuedonymous
      • 5 months ago

      Cross-comparison of TDPs between AMD and Intel tells you basically nothing of value about actual power consumption (and its most important proxy, battery life).

      For example, [url=https://techreport.com/blog/32904/how-much-does-screen-size-matter-in-comparing-ryzen-mobile-and-kaby-lake-r-battery-life<]a 2500U draws real-world more than double an 8250U + MX150 combo[/url<], but by TDP it should be an even match (150W TDP for both the 2500U and 8250U) or a cakewalk for the 2500U (if you add the 10W TDP of the MX150).

    • dragontamer5788
    • 5 months ago

    [quote<]My buddy picked up a machine with a Ryzen 5 3500U and after putting an SSD in it (because, sigh, it came with only a hard drive)[/quote<] Honestly, I prefer this. Dell has long-term contracts so they're unable to offer good prices on SSDs. And since SSDs have dropped like, 50% in price over the last 6 months, this is to a huge benefit to (slightly technical) consumers with a spare screwdriver. If they stick a hard drive in there, then its a mass-market product that everyone can use... and technical folks can easily upgrade. Bonus points: the HP motherboards seem to have a M.2 slot and a SATA slot, so you can get great upgrade potential on those.

      • RAGEPRO
      • 5 months ago

      Yeah, it’s not so bad. Especially when they have (like the HPs you mentioned) an M.2 slot and a 2.5″ bay, so you can keep the HDD and throw in an SSD.

      • cynan
      • 5 months ago

      [quote<]Dell has long-term contracts so they're [b<]unable[/b<] to offer good prices on SSDs[/quote<] Do you work for Dell? Because they play the SSD futures game and don't come out on top at the bottom of the memory demand cycle, they can't [i<]afford[/i<] to sell SSDs at a competitive price? And even more astounding, selling mechanical disks in their place is a service to the customer? If they can't sell virtually commodity products at market competitive prices, they are in the wrong business. And at the very least should sell the PC without either SSD or HDD and not force customers to pay for something they don't necessarily want (or something that is overpriced)

        • dragontamer5788
        • 5 months ago

        Futures contracts cost money too yo. Theta decay, etc. etc. The technical end-customer like you or me can win the game by simply:

        1. If SSD prices go up, just buy a Dell. (leveraging their future contracts and take advantage of a steady price).

        2. If SSD prices go down, we build our own. Dell still had to buy future contracts just in case prices rose, so they’re a disadvantage in this market.

        2018 was a good time to buy pre-built. Individual RAM prices were outrageous, but pre-built computers held steady. 2019 is the opposite situation: RAM and SSD prices have absolutely cratered, so the futures shenanigans that a company like Dell played are wasted.

        At least Dell and HP still allow the end user to self-administer upgrades. Companies like Apple or Lenovo are beginning to hard-solder their equipment to the motherboard. You literally can’t play this game with Apple anymore.

        I prefer to play the game Dell and HP plays. They almost certainly build the low-end assuming that technical users can (and will) upgrade them by hand. And yes, Dell and HP have been known to leave empty M.2 slots ready for the end-user to populate on their own, as well as offering detailed manuals for free on their webpage.

        [quote<]And at the very least should sell the PC without either SSD or HDD and not force customers to pay for something they don't necessarily want (or something that is overpriced)[/quote<] Mass production and SKUs are funny. Its probably cheaper to keep the hard drive in there through the magic of mass production rather than build and support a new low-volume SKU.

          • cynan
          • 5 months ago

          Absolutely. But the market will dictate whether Dell will be able to gouge (i.e., how much more they will be able to charge than current market forces dictate) customers for their deployed memory products or not, regardless of the reason. If Lenovo, HP, etc do it to, then Dell can do it. If not, well, they’re no Apple.

      • Mr Bill
      • 5 months ago

      +3 Agreed and you can put in the brand, model, and size you want.

    • ronch
    • 5 months ago

    So the 3500U runs at 2.1/3.7 while the older 2500U is 2.0/3.6.

    Well, wow. Onwards and upwards!!! 😀

      • gerryg
      • 5 months ago

      The GPU peak clock is also 100Mhz more. The selling point for these Pro chips is the extra security features plus longer-term warranty and availability commitments. No telling if there are other architectural improvements, e.g. supports faster RAM or better processing pipelines, but again not the point for a Pro release.

        • ronch
        • 5 months ago

        These chips will likely be the same as the next wave of regular Ryzen Mobile chips, minus the Pro suffix. Looks like Ryzen Mobile won’t be moving forward much this year. But 5% higher CPU base clock is better than nothing, I suppose.

      • Goty
      • 5 months ago

      Kind of like 90% of the Cascade Lake lineup. At least these come with a superficial process node bump.

      • Spunjji
      • 4 months ago

      I’m interested to see more information about power draw, as performance was already competitive and (in theory) they could be making use of the Zen+’s better voltage characteristics instead of going for outright performance.

      Notebooks based around the original Zen APU were universally pretty terrible for idle power. I’ve only seen one evaluation of a Zen+ mobile chip so far, and even though it was one of the 35W TDP models (3550H) the idle power numbers are a significant improvement compared with a comparable device running the 2700U.

    • Krogoth
    • 5 months ago

    I’m wondering if these SKUs have unbuffered ECC support.

      • RAGEPRO
      • 5 months ago

      Regular old Ryzen [i<]can[/i<] have ECC UDIMM support—I believe it varies based on motherboard firmware?—so I reckon these could too.

        • Krogoth
        • 5 months ago

        Most of them have unofficial ECC UDIMM support and it technically works but they don’t do the “2-bit error detection” system halt.

        That’s why motherboard vendors cannot claim full ECC support which requires the aforementioned function.

          • Waco
          • 5 months ago

          For a consumer machine, just correcting/reporting the errors is all most want/need.

          It was a pain finding a good TR board to run with ECC. I wish AMD would have gone the extra tiny step in making sure it worked for non-EPYC chips, officially.

          Maybe for the 3000 series? 🙂

            • Krogoth
            • 5 months ago

            Right, but from a support/vendor standpoint. The 2-bit error detection halt is required for official ECC memory support. 😉

            • Waco
            • 5 months ago

            Agreed, which is why I hope AMD does better going forward. Advertising 99% of what consumers want and ensuring it’s clear that 2-bit fault halt isn’t supported would be worlds better than what we have now…which is random consumers who are interested trying out various boards and hoping future BIOS updates don’t screw up ECC operation.

        • derFunkenstein
        • 5 months ago

        I thought ECC support, even unofficial support, was missing from APUs entirely? Nothing official on that, though. I just remember reading forum threads like this:
        [url<]https://www.phoronix.com/forums/forum/hardware/processors-memory/1017727-apparently-raven-ridge-apus-don-t-support-ecc[/url<] And Reddit topics like this: [url<]https://www.reddit.com/r/Amd/comments/92jck3/is_anyone_using_ecc_ram_with_their_ryzen_2400g/[/url<]

          • RAGEPRO
          • 5 months ago

          Yeah, apparently so, also per chuckie’s post below. I didn’t know. Mea culpa.

      • chuckula
      • 5 months ago

      Raven Ridge APUs do not have ECC support*, so I’d look to see if AMD lists official support or not.

      * Even if a motherboard lists ECC that might work with a non-APU RyZen that doesn’t mean an APU in the same board supports it.

        • mczak
        • 5 months ago

        Raven Ridge (as in the actual hw) should support ECC – there’s the embedded V1000 cpus which officially support it, which use the same die.
        However, ECC apparently does not work with the ordinary desktop/mobile Raven Ridge APUs for some reason (maybe disabled in AGESA code?), while it does with Summit/Pinnacle Ridge (unofficially). Some motherboards actually do state it should work with the Pro APUs however, like here:
        [url<]https://www.asrock.com/mb/AMD/X470%20Taichi/index.asp#Specification[/url<] "*For Ryzen Series CPUs (Raven Ridge), ECC is only supported with PRO CPUs." I haven't actually seen anyone verifying this, though...

          • stdRaichu
          • 5 months ago

          Good luck trying to find anyone who’s managed to get their hands on one of the Ryzen Pro APU outside of a pre-built box…

            • mczak
            • 5 months ago

            Right, which is why noone has actually verified if these chips indeed would support ECC (I don’t think any of the pre-built boxes using these chips support ECC) – you’d literally have to get one of those pre-built systems using such a cpu, rip the cpu out (obviously this applies to the non-mobile chips only) and try it in one of the boards which state they support ecc for the pro apus.
            So if they support ECC or not might be an interesting technical question, but it’s pretty much moot if none of the pre-configured systems do (regardless if the cpu would) and you can’t actually buy the cpus retail.

    • enixenigma
    • 5 months ago

    “further platform updates”

    Zen 2 update?

    You’d think that AMD would want to fast-track getting their new hotness into mobile (maybe with Navi as well? A man can dream) to get those sweet, sweet design wins.

    • chuckula
    • 5 months ago

    Nice try AMD, but unless you put in at least three chiplets with 16 cores and no graphics these things clearly don’t address any commercial market space that I’ve ever heard of.

      • RAGEPRO
      • 5 months ago

      You struck a nerve with this one chief.

        • chuckula
        • 5 months ago

        It must be paid Intel shills.

        After all, Papermaster got up on stage and said that any product that doesn’t use chiplets is obsolete!

        And any paid shill who disagrees with him should just look at the Radeon VII with its failed non-chiplet design and see that he’s right!

        • BobbinThreadbare
        • 5 months ago

        it’s just not funny tbh

        • Beelzebubba9
        • 5 months ago

        You mean the readership is tired of chuck tilting at the same lame windmill post after post, year after year? He’s just a loser, and that’s fine. But he can learn to keep useless drivel like this to himself.

      • chµck
      • 5 months ago

      i see these working great in NUCs that satisfy the power needs of, i suspect, a large majority of office jobs

Pin It on Pinterest

Share This