Gadgetory


All Cool Mind-blowing Gadgets You Love in One Place

AMD Ryzen, Faster Memory, and the Infinity Fabric

2017-12-11
following the initial release of AMD Rison there was a ton of discussion regards to using higher clocked system ram particularly in games the performance benefits of having higher clock memory was something that we didn't necessarily see an Intel side it's just not now the architecture is made it's not made to benefit from higher clock memory but rise ins architecture is and that's what I want to talk about in this video if it's worth buying really expensive high clock RAM out of the box for the sake of a rise in CPU and a lot of the people who spend you know 400 bucks on an 1800 X are probably gonna buy a higher clock to rise in memory anyway at least a memory that's capable of reaching 3000 or 3,200 maker it's on the rise in platform which is difficult to do we'll talk about that as well but for those who are in the mid-range Rison five-tier should you still consider fast memory and why does it even matter in the first place I talk about all that in this video now when it comes down to it all rise in CPUs on a fabrication level are basically the same there are a few imperfections in some of the cores and that is ultimately differentiates a rise in three cpu from a rise in seven cpu so of the eight cores that are packed into every Rison chip no matter if you have a rise in three up to Horizon seven CPU if four of those aren't doing too well then AMD is gonna just disable and they're just gonna turn them off it's better than making an entirely new fabrication process for a four core chip this way they can save money makes perfect sense I don't blame them for doing that so a before remaining course if these can handle power loads pretty well and the the performance degradation isn't too substantial then these might be arisin five 1400s or 1500 X's if they aren't doing too well with handling a lot of power and the schedulers aren't doing that great a job at increasing the performance level then they might be rising three CPUs with multi-threading and disabled so it really comes down to what AMD deems as acceptable on both power delivery and execution levels now the difference between AMD and Intel because Intel does this too with bidding is that AMD decides well if these four cores aren't going to cut it we're gonna disable two of the cores in one cc X which is a four core cluster in a Rison CPU and two cores in the other CC X which is the other cluster of horizon CPU every rise in CPU has to CC x's per die the red Ripper has four which is why you can get up to 16 cores in those CPUs maybe this diary shot will help explain things a bit better so if two cores are activated in one cc X and two are activated in the other there has to be some sort of efficient way for them both to communicate because these are essentially two separate dies they might not be that physically but each core cluster has an own set of resources including l3 cache which means that when you're in especially intense situations on a computational level and they have to share information and process information simultaneously and exchange resources there has to be an efficient Highway a way for data to be transferred super quick between the cores to eliminate that latency this highway is called the Infinity fabric as Tom's Hardware puts it the large amount of data flowing through this pathway requires a lot of scheduling magic to ensure a high quality of service it's also logical to assume that these six and four core models benefit from less cross CCX traffic compared to the eight core models so they're actually hinting at here is the dual CCX design is a blessing in a way and a curse it's a blessing for AMD from a financial standpoint because all they have to do is slap multiple CC exes into the same chip and just run infinity fabric all between them which saves them money because they can control how many CC exes are in each dial without completely redoing the fabrication process but the curse involved is the latency involved between the CC X data transfers so the Infinity fabric itself is not the most efficient means by which data is transferred from one CC X to another this is where the faster memory comes into play it's but this whole video is about why rise and benefits from faster memory we hear people say that maybe they don't know what they're talking about but they are correct in a sense it really depends on the number of cores enabled per cc X but ultimately Rison will benefit more from faster memory than Intel will and the reason why is because the Infinity fabrics speed this the rate at which it can transfer data is directly tied to the frequency of your system memory Intel is independent on the same variable because their consumer-grade CPUs are reliant on the ring bus design two rings share information between let's say 4 cores and basically you have less traffic being dispatched between cores so the the congestion isn't as great as it would be on an 8-core Rison cpu because then you have four cores trying to share information down let's say a single pathway with the Infinity fabric so there can be quite a bit of congestion and lag I shouldn't say lag it's more or less a gamer term but latency it's the delay in data transfer and we see a huge difference between the rain bus design with Intel a comparable Intel CPU and the Infinity fabric whereas cores on an i7 7 700 K rely on the latency roughly between 30 and 40 nanoseconds cross quartic or latency on a cc x-ray for Horizon is somewhere in the realm of 200 nanoseconds almost 10 times the latency just because data has to be transferred across an infinity fabric not to be frank we're talking nanoseconds here not even milliseconds but it does add up over time as huge computational workloads bottleneck that infinity fabric you can imagine how things get pretty backed up in the long run and you might even be able to tell a difference in things like games and also in render times if you do have heavy workloads being pushed to rise in CPUs they might not be able to handle them as well with that lower clocked system Ram so there you have it yes AMD rice and CPUs do benefit from higher clock memory it's not a myth and the reason why is because AMD employs infinity fabric which is directly dependent on the speed of your RAM now getting too much into detail that's really all you need to know something else worth noting is that rise in 3cp is because they only have two cores activated per CCX aren't going to benefit as much from the higher clocked Ram only because the data transfer rates between CC X's are going to be as high because only two cores per CC X are actually sending information whereas in horizon 7 1700 700 X or 800 X CPU get four cores on each side sending a ton of information things can get pretty congested in there for more info on this maybe you'd like to read an article about the Infinity fabric I invite you to check out the link at the top this video's description sometimes it helps to just read something over and over until it clicks each it takes me 5 or 6 times I'm like I'm like you're bringing the sentence until till something clicks you get that light bulb go off now ok I get it now it's harder to do that with a videocassette to keep hearing me say it over and over again it's less annoying when you're doing it yourself in your head and invite you to check that link out by the way it's linked to Tom's hardware and they have a great article breaking down the Infinity fabric and why it behaves the way it does also is a pretty cool benchmarks in there to back up the claims made in this video if you like this video be sure to give it a thumbs up I appreciate it thumbs down for the opposite click to subscribe but if you have any stay tuned for more content like this this is science studio thanks for learning with us
We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites.