In the lab: AMD’s Ryzen 5 1600X and Ryzen 5 1500X CPUs

AMD's Ryzen 5 processors don't launch until April 11, but a large box has made its way to my doorstep with a whole bunch of goodies inside. We can't talk about how Ryzen 5 CPUs perform yet, but we can show you what we'll be using to test them.

It's a box.

There's a pair of CPUs in it. The Ryzen 5 1500X has a Wraith Spire cooler in the box with it, hence the much larger package.

Here's the Wraith Spire in its non-RGB LED form, next to a Haswell or similar-era Intel stock cooler.

AMD also sent along the Wraith Max RGB LED-illuminated cooler. You can't buy this cooler or get it with a CPU in stores, but prebuilt Ryzen systems with fancier CPUs inside will probably come with this heatsink installed.

This is Gigabyte's handsome AB350-Gaming 3 motherboard, our B350 pick for AM4 CPUs from our most recent System Guide.

AMD also sent along a kit of this outrageously-tall GeIL EVO X DDR4-3200 RGB LED-illuminated RAM. Critically, these are DDR4-3200 single-rank DIMMs rated for 16-16-16-36 timings. Ryzen CPUs can't reach those high memory speeds with dual-rank DIMMs right now.

Seriously, look at how big these things are. I'll probably be seeing a lot of them as I test this Ryzen 5 duo. Stay tuned for our full review April 11.

Comments closed
    • spiritwalker2222
    • 2 years ago

    When are the 3 series coming out? My next PC will be a HTPC for the garage.

    • psuedonymous
    • 2 years ago

    Blimey, double the HSF size (slightly thicker fan and triple-thickness of actual heatsink), but for the ‘same’ 65W TDP? Either that’s intended to be the most silent stock cooler ever, or the DP calculation disparity for Ryzen 5 is going to be as large as for Ryzen 7 (e.g. drawing the same actual power as a ‘140W’ Broadwell part when supposedly a 95W TDP).

      • just brew it!
      • 2 years ago

      …or perhaps they’re actually supplying a stock cooler which is usable for moderate overclocking. We can hope.

    • Unknown-Error
    • 2 years ago

    In gaming you can expect even worse performance.
    BDW-E vs RyZen: gaming tests

    [url<]https://www.purepc.pl/procesory/amd_ryzen_7_1800x_vs_intel_core_i7_6900k_analiza_wydajnosci[/url<] BDW-E 6900K is 40% to 60% faster than the higher clocked 1800X. Zen is a gaming failure. [url<]https://www.purepc.pl/procesory/amd_ryzen_7_1800x_vs_intel_core_i7_6900k_analiza_wydajnosci?page=0,9[/url<]

      • raddude9
      • 2 years ago

      Why are you linking to a very questionable Polish Ryzen review instead of just pointing people to this very site, in particular, the conclusion page:
      [url<]https://techreport.com/review/31366/amd-ryzen-7-1800x-ryzen-7-1700x-and-ryzen-7-1700-cpus-reviewed/13[/url<] Where you can clearly see that the gaming performance of Ryzen is only maybe 10% behind BDW-E.

        • freebird
        • 2 years ago

        Because they are a paid shill probably…

    • Laykun
    • 2 years ago

    Now these are going to be the interesting chips for gamers, if that 6 core can hold it’s own against other lowish i7s or i5s we might have a winner. If the disabled cores can be unlocked in a CPU lottery like fashion then that’d be even better.

      • DancinJack
      • 2 years ago

      doubt it

      • DoomGuy64
      • 2 years ago

      The 6 core paired with some 3200 ddr should be a pretty competent gaming cpu. Otherwise, the CCX issue will be worse than the 8 core chip. That said, intel doesn’t offer anything compelling against the 1600X, so I can see it being a relatively popular CPU with people looking to upgrade from older i5’s on the cheap. Quad cores are getting too dated to continue buying, even if single threaded performance is slightly better, especially when games using 6+ cores or hyper-threading will perform better on the 1600X. The clockspeed trade off should be worth it, especially if you OC and pair with fast memory.

    • odizzido
    • 2 years ago

    To the people at TR running these benchmarks. You should check out this video
    [url<]https://www.youtube.com/watch?v=0tfTZjugDeg[/url<] TL;DW - check out the graph here [url<]https://youtu.be/0tfTZjugDeg?t=15m7s[/url<]

    • WaltC
    • 2 years ago

    What I read a few days ago about the Wraith MAX RGB cooler is that you will be able to buy Ryzen cpus with it–or without it–as AMD will be offering both skus. So you should be able to purchase a boxed Ryzen Kit with this fan. But that is *all* I have read about it…;) So don’t know for sure. For gaming, the six-core sounds very nice, for me.

    • Ninjitsu
    • 2 years ago

    Erm what is this single/dual rank thing? As in, what does that mean?

    I have a legitreviews tab open which goes into single vs dual rank perf, doesn’t seem to actually explain what “rank” means…

    As an aside, I wish they’d give the 6 core part a different series name than the quad cores – calling them both Ryzen 5 is really confusing.

      • Jeff Kampman
      • 2 years ago

      Single-sided (or single-rank) == the memory chips are on only one side of the DIMM

      Double-sided (or double-rank) == the memory chips are on both sides of the DIMM

      A bit more detail: [url<]https://en.wikipedia.org/wiki/Memory_rank[/url<]

        • Ninjitsu
        • 2 years ago

        Ahhh. LOL. I didn’t know that’s what they called them, I knew them simply by some other name I think. Single/double sided or something.

        Thanks for explaining it!

        • D@ Br@b($)!
        • 2 years ago

        No. It is a common misunderstanding that double sided and dual rank are the same thing.
        As explained in the link that you provided and the following one,

        [url<]https://en.wikipedia.org/wiki/DIMM#Ranking[/url<] From another link which I already posted in this thread, der8auer(on overclocking guide): "Ranks are administration units which are formed out of a set of memory chips. Usually 8 IC's form one rank. Information stored in RAM cells are just charged capacitors and due to the low capacity it will lose the information after few nano seconds. To keep the information stored it’s necessary to perform a refresh cycle every few nano seconds. Rank interleaving is a DDR3 feature which allows to perform a refresh operation in one rank and a access operation in the other rank using dual-ranked memory modules. Single-ranked memory modules can only perform one operation at the same time which will lead to a slower performance because you have to wait one cycle until you can access the information stored on your stick." Double ranked modules generally perform slightly better and will strain the memory controller more. But, LOL, with DDR4, which is the subject matter in the article, I dunno, der8auer: "I will also publish a DDR4 guide about single/dual rank and other myths. Some stuff is the same like Myth #1 but single and dual ranked is only a small issue because DDR4 works different. " Ninjitsu, can you post a link to the Legitreviews tab. Was it about DDR3 or DDR4

          • blahsaysblah
          • 2 years ago

          [url=http://www.legitreviews.com/amd-ryzen-single-rank-versus-dual-rank-ddr4-memory-performance_192960<]AMD Ryzen – Single-Rank Versus Dual-Rank DDR4 Memory Performance[/url<]

            • D@ Br@b($)!
            • 2 years ago

            Tnx for the link.
            Apparently single rank memory is marginally faster overall on Ryzen, “au contraire” to Haswell-E:
            [url<]http://www.corsair.com/en-us/blog/2014/september/ddr3_vs_ddr4_synthetic[/url<] where dual rank is marginally faster.

          • Ninjitsu
          • 2 years ago

          DDR4 – [url<]http://www.legitreviews.com/amd-ryzen-single-rank-versus-dual-rank-ddr4-memory-performance_192960[/url<] EDIT: oh i see blahsaysblah linked to it already. oops!

          • freebird
          • 2 years ago

          -Good info…

          The important point to remember is that “the more Ranks, the more load on the Memory controller”, which can force you to run more ranks at a lower speed than less ranks. This is why you now see LR-DIMMS in Server Memory. Load Reduced Memory Modules have an on chip buffer to reduce the electrical load.

          RDRAM (RAMBUS) went a step further and put some type of memory controller on the Memory Module (which also helped lead to its downfall)

      • albundy
      • 2 years ago

      its another bs invention to make things way more difficult than they should be. no etailer lists this spec, and its quite hard to find this info on the manufacturer website.

        • Ninjitsu
        • 2 years ago

        kinda, they usually have the image on the spec sheet, or number of modules per stick, or density. Admittedly a roundabout way of figuring it out.

    • 1sh
    • 2 years ago

    These are not native quad and six core dies but octo core dies with 2 or 4 cores disabled.
    Correct me if I am wrong.

    I was initially expecting higher clocked quad and six cores…

      • deruberhanyok
      • 2 years ago

      correct:

      [url<]http://www.legitreviews.com/amd-ryzen-5-cores-are-disabled-in-symmetrical-pairs_192827[/url<]

      • just brew it!
      • 2 years ago

      Well, you’ll probably have a bit more thermal headroom at least, since you’ll be dissipating less heat through the same amount of surface area. This may help with OCing.

        • 1sh
        • 2 years ago

        Hmm, I am assuming the 1500x can hit 4.5ghz if the octo core 1700 can hit 3.7 to 4.1ghz.
        I would like to see how the power consumption increases when OCing the 1500x.

          • just brew it!
          • 2 years ago

          Probably not a good assumption. Thermals aren’t the only thing that cap your maximum stable clock speed.

          • NoOne ButMe
          • 2 years ago

          [url<]http://i.imgur.com/8Rch6JF.png[/url<] This should hold true for cut down Ryzen CPUs. A new stepping/revision could push the critical points and clockspeeds up. Hopefully Raven Ridge brings some improvement.

          • DancinJack
          • 2 years ago

          nope

          • ptsant
          • 2 years ago

          The CPU is not limited by thermals or current. The sweet spot for Zen is around 3.0-3.5GHz. Even a single-core Zen would not go much further than 4.1-4.2.

          Wait for next revision if you care about OC or single-threaded performance.

    • derFunkenstein
    • 2 years ago

    You should like the AB350-Gaming 3. I’ve found it to be exceptionally stable and able to deal with my Corsair DDR4-3000 RAM at 2933 speeds (since there’s no multiplier for an even 3000). And that’s with dual-rank DIMMs, since they’re 16GB a piece. The Wraith Spire will likely cool everything but the most extreme overclock, too. I’ve been fiddling around and I’m up to 3.8GHz at 1.325 and the temps stay in the mid-70s under a Prime95 load.

    And whoa, they’re giving you, like…10 days to do the review (assuming a reasonably early NDA lift time). That’s pretty outstanding.

      • D@ Br@b($)!
      • 2 years ago

      Must be a really bad 1800X you have there,
      apart from the memory controller.

        • derFunkenstein
        • 2 years ago

        It’s a 1700 non-X. Instead of a single-core turbo of 3.7 it’s an all-core turbo of 3.8, which is quite a nice bump.

          • D@ Br@b($)!
          • 2 years ago

          Ah, ok.
          Does the memory overclock affect the CPU overclock ability, or vice versa, like it does with Haswell?

            • derFunkenstein
            • 2 years ago

            Doesn’t seem to. I swapped motherboards due to an annoying POST issue that affected the B350 Tomahawk. I’ve got that AB350-Gaming 3 now and that went away. The OC I have set up is the same as I had before, when that board would only get the RAM up to 2666 with manual settings. The Gigabyte board seems capable of handling both the SMP profile and POSTing every time.

            A beta BIOS has become available for the MSI that supposedly fixes the issues I had (which affected me even at stock and with the memory running at 2133). So it might be fine now.

            Either way I’m not willing to turn the screws on it until NZXT ships me an AM4 bracket for the Kraken X31 in that system. I got an email from them yesterday saying it’s expected to ship early next week.

            • D@ Br@b($)!
            • 2 years ago

            Well, good luck with your further overclocking efforts. Maybe you will end up with a 1800X after all. πŸ˜‰

        • ptsant
        • 2 years ago

        He is also using a conservative voltage. People regularly pump 1.45V to achieve 3.9-4.0GHz. I suppose he should be able to hit at least 3.9-3.95 with 1.4V. More than 1.45V is not good for the longevity of the CPU according to AMD.

          • anubis44
          • 2 years ago

          I found that 1.375v-1.3825v was the sweetspot for overclocking my Ryzen 7 1700 on the Gigabyte AX370 Gaming 5 I had for a few days. I was able to clock it to 3.95GHz stable (tested with CPU-Z CPU Test function) using the stock AMD Wraith cooler, though I didn’t try Prime 95. I loved the Gaming 5’s stability, and it was able to clock the G.Skill Trident Z 32GB (2×16) 3200MHz CL14 ram to 2933MHz, but I could not push it any higher, so I decided to take the board and the ram back to the store for a credit. Waiting on either the Gigabyte AX370 Gaming 7 or the Asus Crosshair VI Hero, as both of those have external BCLK generators, which seems to allow higher memory speeds than board without them, but I’m sitting tight until I see which board(s) consistently hit the highest memory speeds, and with which RAM before I buy something else.

    • just brew it!
    • 2 years ago

    Serously, WTF is with those DIMMs? That’s gonna restrict airflow in the area of the CPU socket!

      • D@ Br@b($)!
      • 2 years ago

      Yes and that is better.
      The Wraith cooler sucks it in from above and the DIMM’s force more of the air to the voltage regulators.

        • just brew it!
        • 2 years ago

        …and then the DIMMs deflect that hot air back towards the top of the HSF, causing it to get sucked back in again.

          • D@ Br@b($)!
          • 2 years ago

          Yeah that too… πŸ™‚

          Edit: added a nose

    • HERETIC
    • 2 years ago

    What might be interesting is if you also test your 1800 with one cluster/module disabled-
    see what effect the fabric/cache has.compare 4-0 to 2-2.
    Being realistic you probably already feel there’s not enough hours in a day…………..

      • jessterman21
      • 2 years ago

      [url<]https://youtu.be/Rhj6CvBnwNk[/url<] Turns out there's not much of a difference...

    • TwoEars
    • 2 years ago

    This is the really interesting launch, looking forward to it. With the right price the 1500X is the cpu 80% of regular consumers should be interested in.

    • sophisticles
    • 2 years ago

    Very elaborate April Fool’s day write up, not bad. Of course the dead giveaway is the supposed DDR4 3200 ram…

    • ronch
    • 2 years ago

    Are you guys planning to put these inside the Fur E enclosure?

      • derFunkenstein
      • 2 years ago

      The Fur E is an mITX enclosure. Good luck shoving a full-sized ATX board in there. :p

        • just brew it!
        • 2 years ago

        Sounds like a job for Mr. Hacksaw to me. πŸ˜‰

          • ozzuneoj
          • 2 years ago

          … incidentally, that is the name of my cat.

    • Unknown-Error
    • 2 years ago

    April Fools’ Day?

      • kuttan
      • 2 years ago

      Bad wit.

    • Bumper
    • 2 years ago

    Im interested in the 1500x. Glad you are testing.

    • Firestarter
    • 2 years ago

    well that RAM is not going to be very representative of the kind of system these CPUs will live in. Whoever is going to shell out for single-rank DDR4-3200 is not going to settle for a Ryzen 1600X

      • DancinJack
      • 2 years ago

      That’s a pretty big assumption. What if someone bought that RAM a year ago when it wasn’t so dang expensive? I’m sure there are plenty of AMD buyers shelling out for good RAM, especially given the fact that Ryzen excels with higher speed modules.

        • Welch
        • 2 years ago

        Exactly. For instance, maybe I want the best of gaming and multitasking. So 2 extra cores help with my streaming!

        I’d rather saved money buying an R5 1600x at $249 vs an R7 1700 at $329. Take the difference and trade those 2 cores for either a bit better memory for the infinity fabrics sake or a better GPU. You have to imagine that the higher end R5 series is meant to be competition to Intel’s i7 for gamer/streamers/content creators (YouTube videos). It’s not by any means a slacker just because something is above it which is clearly now designed for enthusiast professionals.

      • synthtel2
      • 2 years ago

      Me? 1700 non-X and some overkill-fast RAM is the current plan. (Whatever my IMC can handle, I want it to be handling.) I have uses for which having eight cores means I might as well not even worry about core or RAM clocks, and I have uses which I suspect will care just as much about RAM BW and latency as single-threaded performance once that many threads are available.

        • Firestarter
        • 2 years ago

        yeah, 8 overclocked cores would be worth it, but 6 or 4? We already know Ryzen won’t clock much beyond 4.0ghz no matter what you do, and judging from the released clock speeds we shouldn’t expect the 4 or 6 versions to be much different. When the price difference between fast RAM and regular RAM is the same as between your current CPU of choice and basically the same with 2 more cores, it doesn’t make much sense anymore to go with the faster RAM instead of the [s<]faster[/s<]wider CPU. Especially considering how modern games will be more likely to take advantage of those 2 extra cores don't get me wrong, I'd go all out and get the fastest RAM with octa-core Ryzen too, or alternatively with a i7-7700K. But I'd sooner buy a Ryzen 7 1700 with slower RAM than a Ryzen 5 1600X with fast RAM, same as I'd rather have a i7-7700K with slower RAM than a i5-7500 with fast RAM

          • Welch
          • 2 years ago

          But at that rate your trading the single threaded performance for less capable multi threaded performance. Unless you benefit much more from those extra 2 cores in the apps than you do from single threaded. Like you said, put the faster in either anyway.

          As for overclockability, I don’t think we have seen this clear picture yet. The arch may allow for higher clocks but we have only have seen 8 cores trying to be OC’d. AMD does 3+3 and 2+2 on their R5 chips. I’d imagine they were smart enough to find the weakest core on each CCX that can’t clock as high and disable it. Unless Ryzen just hit it’s per core theoretical cap already and the arch doesn’t have any more headroom.

          That would also mean that sub 4 core Ryzen performance will be junk in real world apps other than surfing. Would explain the double stuffed offerings for $100+ less compared to Intel’s offerings.

          • synthtel2
          • 2 years ago

          If my budget were tighter, I’d drop to a 1600 non-X and keep the fast RAM, though I’d be a bit more aggressive with core clocks in that scenario.

          A key factor is that I wouldn’t buy any higher-end chip of a given core count, since I’d just overclock to match them anyway. That makes the 6C <-> 8C price gap ($110) a lot bigger than that between RAM of different performance levels (at least at 2x8G – ~$50). For my uses, I’m pretty sure 6C + fast RAM for $380 is a nicer option than 8C + meh RAM for $440. For someone else’s uses, maybe not.

            • Firestarter
            • 2 years ago

            16GB for a 6 or 8 core CPU in 2017 is scraping the bottom of the barrel. At a memory capacity fitting for a 8 core CPU these days, the price difference between a slow pair and a fast pair of single rank DDR4 DIMMs is a lot closer to the gap between these CPUs

            • synthtel2
            • 2 years ago

            Some workloads have working sets that have to be mostly duplicated for each thread, but mine that are like that don’t have huge memory footprints in the first place. Just because I’m getting 8x the threads doesn’t mean my memory requirements are going to increase much, and I get by very well on 8GB presently.

            My OS doesn’t use much and Firefox isn’t ridiculous, so all but ~2GB tends to be available for whatever my foreground task is at the moment. Swapping to a compressed ramdisk works really well even with one thread on a G3258, and is going to get even better with tons of threads – if I have to swap once in a blue moon, it isn’t the end of the world. Only once has 8GB felt seriously limiting, and that wouldn’t have felt limiting at all if I’d had more threads to throw at memory compression (manipulation of an extremely large but very compressible image).

            Plenty of people have need for 32 GB and plenty don’t. At this level, nobody’s workload is normal, but I got curious about the ratios and made a [url=https://techreport.com/forums/viewtopic.php?f=36&t=119433<]poll[/url<].

            • Firestarter
            • 2 years ago

            interesting, it seems I’m in the minority, 16GB apparently is still good enough for many people

            • LostCat
            • 2 years ago

            8GB would still be good enough for me if it weren’t for a few higher end games starting to choke on it. I don’t think I’ll be going past 16 for a very long time.

      • Welch
      • 2 years ago

      Hehehe, speak for thine self brother gerbil.

      • anotherengineer
      • 2 years ago

      Fast usually is better, but it’s good to know where that diminishing returns point is.

      [url<]https://www.techpowerup.com/reviews/AMD/Ryzen_Memory_Analysis/[/url<]

      • D@ Br@b($)!
      • 2 years ago

      Buying good/fast RAM is always a smart thing to do. The RAM will probably last you a few builds or at least a CPU upgrade.
      AMD will improve the memory controller or else you can always switch to Blue.
      There are of course the diminishing returns, like anotherengineer pointed out, but there are also the bragging rights πŸ™‚

        • Welch
        • 2 years ago

        I’d typically agree, but in light of the new DDR5 announcement that is expected in the next 2 years… I doubt it will last that many builds.

          • D@ Br@b($)!
          • 2 years ago

          Yes JEDEC announced 2 days ago that development of DDR5 standards is moving forward rapidly. Publication is forecasted for 2018 instead of 2020.
          BUT.
          It will take some more years for it to be able to be paired with suitable CPU’s.

          [url<]http://finance.yahoo.com/news/jedec-ddr5-nvdimm-p-standards-184100916.html[/url<] [url<]https://arstechnica.com/gadgets/2017/03/next-generation-ddr5-ram-will-double-the-speed-of-ddr4-in-2018/[/url<]

            • derFunkenstein
            • 2 years ago

            This is all true, and add to that the fact that you won’t see DDR5 on mainstream platforms right away. It’ll probably be an Intel-first on some future platform.

            • Welch
            • 2 years ago

            I’m not sure I buy that really. I think we will see it in platforms sooner than previous iterations. The reason being was Intel was the only major player through DDR3 and a good chunk of DDR2. DDR4 is the first time since DDR that we have had them on a relatively equal playing field.

            The heat is on, and any competitive edge that one can gain on the other will be taken. Look at GPU memory lately, talk about getting dizzy for anyone who hasn’t kept up, GDDR5, GDDR5x, HBM, HBM2. They are bouncing everywhere trying to strike an advantage over each other. Within 2-3 years we will see the newest CPUs using GDDR5 I’m betting. This is easily the life span (4 years according to AMD) on Ryzen and would be perfect timing for Intel to be retiring their Core i architecture after over a decade of use.

            • D@ Br@b($)!
            • 2 years ago

            LOL
            Leaning back and overseeing this part of the discussion I think we agree on the lifespan of DDR4.
            In that time frame I bet AMD will present us some new CPU’s with better IMC and overclock ability on their AM4 platform, barring they don’t pull of a AM4+. πŸ˜‰
            So I stand with my point. Adding ;-). I don’t mean so much faster RAM it has the halo status and matching price.
            (Checking prices in my region)
            Uehh,…. damn. Which the Geil stix mentioned above actually are at this moment. They have a price premium of 50% over 24C12(although without bling) and they sure as hell won’t preform 50% better(in real life)
            Al things said. I would still buy some 3200C16, trading in some beers for water and a few delivered pizza’s for a home cooked meal(or two), being better of in two ways. πŸ™‚

            PS, I think the Team Group Night Hawk modules look really cool. πŸ™‚

          • Umbral
          • 2 years ago

          DDR4 is not going away anytime soon. If only I’d had the foresight to buy a set or two last fall.

      • ptsant
      • 2 years ago

      I used to always buy the best quality RAM and mid-range CPUs. For my Ryzen build I chose capacity over speed, so I got a 2x16GB dual-rank setup. I need at least 32GB and may upgrade at some point to 64GB.

    • blahsaysblah
    • 2 years ago

    Is there any advantage for dual rank DIMMs? or is it just a cost savings to use older/lower capacity parts? All things equal, if you had choice and it wasnt too much different in cost, should you try to get DIMMs with chips on only one side/less chips?

      • Welch
      • 2 years ago

      Single rank means your dimms are all on one side. Which means your DIMMs are going to be twice the capacity. Yes these dense DIMMs are more expensive. However they are faster as far as latency. You have multiple timings, 4 common, although there are actually more. These are ratings in ns of how long it takes the memory to do different things, for instance access one rank, row, or refresh a row or rank. If you have dual rank, it adds latency when having to go between the two ranks. Also sometimes single ranked memory can use less power, making it attractive for overclocking or registered RAM for servers.

      There is most likely an easier way to explain this.

        • synthtel2
        • 2 years ago

        Latency ratings are in cycles, not ns, so 3200 CL15 has lower absolute latency than 2933 CL15.

        I’ve generally interpreted single-rank’s OC superiority to be due to fewer things (less capacitance) hanging off each line, but I don’t know how true that is.

          • Welch
          • 2 years ago

          true latency (ns) = clock cycle time (ns) x number of clock cycles (CL)

          Stolen from Crucial website on the subject. Actually gives more detail that I even wondered about with the advent of the newer DDR standards.

          [url<]http://www.crucial.com/usa/en/memory-performance-speed-latency[/url<]

          • Welch
          • 2 years ago

          Yeah, I’d agree. I think their overclockability may have something to do with less power required overall and probably similar to multiple cores on a CPU, less chips to worry about whether they can OC. If you’ve got twice the chips, you drastically increase your odds one or more can’t keep up with the pack.

          Not sure if they bin memory the same was CPU chips are for single vs dual ranked. Hmmm.

      • yuhong
      • 2 years ago

      The way it generally works is that when for example 8Gbit DDR4 chips first appear, dual rank modules appear first then eventually as it gets closer to crossover single rank modules appear to replace 4Gbit dual rank modules. The fun thing is that the price for 4Gbit/8Gbit DDR4 is still going back and forth on DRAMeXchange.

      • D@ Br@b($)!
      • 2 years ago

      There is with DDR3. But speed and latency, of course, also matter.

      [url<]http://overclocking.guide/ddr3-ram-myths-enlightened/[/url<] Duckduckgo yourself for DDR4

        • blahsaysblah
        • 2 years ago

        Thanks for that link. Very goid background.

      • Krogoth
      • 2 years ago

      Dual rank DIMMs have greater capacity at the cost of sightly increased latency and more power consumption/strain on memory controller.

      Memory capacity is king with memory bandwidth following in with a distant second.

      • just brew it!
      • 2 years ago

      Aside from the points others have already mentioned, all else being equal dual-rank may have a slight edge in workloads with highly random memory access patterns since they allow more banks to be kept open at the same time. Typically this will be overshadowed by the need to use looser timings though.

        • blahsaysblah
        • 2 years ago

        Thanks, thats exactly what i was curious about. If more chips were better for high concurrent memory usage scenarios, like multiple VMs.

    • Yan
    • 2 years ago

    Illuminated RAM? πŸ™

      • caconym
      • 2 years ago

      I mean, it’s right there in the abbreviation though: Rgb Accent Module.

      • maxxcool
      • 2 years ago

      Has it see the light ?

      • freebird
      • 2 years ago

      It’s a conspiracy… they are called illuminati RAM.

    • thedosbox
    • 2 years ago

    Where’s the cat?

    WE DEMAND CAT!

      • LostCat
      • 2 years ago

      Busy. Doing…stuff.

        • UberGerbil
        • 2 years ago

        Stuff like… [url=https://techreport.com/news/31678/mad-catz-files-for-chapter-7-bankruptcy-and-shuts-down<]bankruptcy paperwork[/url<].

      • UberGerbil
      • 2 years ago
    • odizzido
    • 2 years ago

    This is the release I’ve been waiting for. I will be pretty interested in seeing the performance comparison to the old benchmarks with all of the software changes that have been going around.

Pin It on Pinterest

Share This